69541                   What Have Been the Impacts of World Bank Community-Driven Development Programs? CDD Impact Evaluation Review and Operational and Research Implications Susan Wong The World Bank Social Development Department Sustainable Development Network May 2012     ©  The World Bank  Social Development  1818 H Street, NW  Washington, DC  20433    Manufactured in the United States of America  First published May 2012    Fax: 202â€?522â€?3247  Eâ€?mail: asksocial@worldbank.org  Online: www.worldbank.org/socialdevelopment    The  findings,  interpretations  and  conclusions  herein  are  those  of  the  author(s)  and  do  not  necessarily  reflect  the  views  of  the  International  Bank  for  Reconstruction  and  Development/  World  Bank  and  its  affiliated  organizations,  or  its  Executive  Directors,  or  the  governments  they represent. The World Bank does not guarantee the accuracy of the  data  included  in  this  publication  and  accepts  no  responsibility  for  any  consequence of their use.    To  request  copies  of  the  paper  or  for  more  information  on  the  series,  please contact the Social Development Department        Printed on Recycled Paper        Table of Contents  Acknowledgments ................................................................................................................................................................... ii  List of Acronyms and Abbreviations .............................................................................................................................. iii  Executive Summary ............................................................................................................................................................... iv  I.  Introduction .......................................................................................................................................................................... 1  II.  Objectives of the Study .................................................................................................................................................... 2  III.  Study Methodology ......................................................................................................................................................... 4  IV.  Conceptual Framework behind CDD Interventions .......................................................................................... 8  V.  Country Contexts ............................................................................................................................................................. 11  VI.  Impact Evaluation Results and Findings .............................................................................................................. 15  VII.  Cost Analysis .................................................................................................................................................................. 37  VIII.  Major Factors Affecting Implementation and Outcomes ........................................................................... 41  IX.  Discussion of Operational and Research Implications ................................................................................... 46  X.  Concluding Remarks....................................................................................................................................................... 52  ANNEX A.  Summary of Projects and Impact Evaluations .................................................................................... 54  . ANNEX B.  Individual CDD Project Descriptions ...................................................................................................... 59  Bibliography ............................................................................................................................................................................ 75    Figure  Figure 1. Illustrative CDD Results Chain ........................................................................................................................ 8    Tables  Table 1. World Bank CDD Programs Covered in this Study ................................................................................... 6  Table 2.  Country Contexts and Origins of the Programs ...................................................................................... 12  Table 3.  Summary of Impact Evaluation Results ..................................................................................................... 34  . Table 4.  Block Grant Per Capita Allocation Amounts by Project ...................................................................... 37  Table 5. Levels of Community Contribution by Project ......................................................................................... 38        i  Acknowledgments    Many colleagues have provided insights and advice invaluable in writing this paper and refining the analysis  and  operational  and  research  implications.    This  paper  has  benefited  from  several  colleagues’  feedback  on  earlier drafts as well as overall discussions regarding CDD operations, impact evaluation, and research.  The  author is especially grateful for comments from Scott Guggenheim, Julien Labonne, Andrew Beath, Ben Olken,  Helene  Grandvoinnet,  Janmejay  Singh,  Richard  Calkins,  Kaori  Oshima,  Vijayendra  Rao,  Ghazala  Mansuri,  Parmesh Shah, Daniel Owen, Rodrigo Serranoâ€?Berthet, Susanne Holste, John Elder, Cecilia Costella, Lily Hoo,  and Myrtle Diachok.  The author also thanks the formal peer reviewers for their useful insights on this paper:   John Elder, Jorge Munoz, Andrew Norton, and Lant Pritchett.    Kaori Oshima provided valuable research support for this paper, and Elizabeth Acul provided administrative  assistance.  Robert Livernash assisted in editing.  ii    List of Acronyms and Abbreviations    ASIF    Armenian Social Investment Fund  CDC    Community Development Council in Afghanistan  CDD    Communityâ€?driven development  DIME    Development Impact Evaluation Initiative of the World Bank  DPIP    District Poverty Initiatives Program  FISE    Nicaragua Emergency Social Investment Fund (Fondo de Inversiones Social de Emergencia)   FONCODES  National  Fund  for  Social  Compensation  and  Development  in  Peru  (Fondo  Nacional  de  Compensacion Desarrollo Social)   GAM    Free Aceh Movement    GoBiFo    “Forge Aheadâ€? in Sierra Leone’s national lingua franca, Krio  ICR    Implementation completion report  IE    Impact evaluation  KALAHIâ€?CIDSS  Kabit­bisig Laban Sa Kahirapan –Comprehensive and Integrated Delivery of Social Services  program   KDP    Kecamatan (subdistrict) Development Program  KDPâ€?BRA  Kecamatan Development Program/Aceh Reintegration Body (Badan Reintregrasi Aceh)  MILF    Moro Islamic Liberation Front  NGO    Nongovernmental organization   NPA    New People’s Army in the Philippines  NSP    National Solidarity Program in Afghanistan  O&M    Operations and maintenance  PAD    Project appraisal document  PAF    Poverty Alleviation Fund in Nepal  PNIR  National  Rural  Infrastructure  Program  (Programme  National  d’Infrastructures  Rurales)  in  Senegal  PNPMâ€?Generasi  National  Program  for  Community  Empowerment  –  Smart  and  Healthy  Generation    in  Indonesia (Program Nasional Pemberdayaan Masyarakat Generasi Sehat dan Cerdas)  PNPMâ€?Rural  National  Program  for  Community  Empowerment  in  Rural  Areas  in  Indonesia  (Program  Nasional Pemberdayaan Masyarakat)  SF    Social funds  SHG    Selfâ€?help groups in India  SIF    Social investment funds  TASAF    Tanzania Social Action Fund (TASAF)  UPP    Urban Poverty Program in Indonesia  WB    The World Bank  Note:  All dollar amounts are U.S. dollars unless otherwise indicated.    iii    Executive Summary     Governments  and  the  World  Bank  have  very  few  tools  to  reach  large  numbers  of  the  poor  directly,  particularly  in  the  context  of  weak  or  fragile  states,  in  postâ€?conflict  and  postâ€?disaster  environments,  or  in  areas  with  poor  track  records  of  service delivery  within  the  bureaucracy.    Communityâ€?driven  development  (CDD) potentially constitutes an important approach in the repertoire of development interventions because  it is designed to place less stress on government line agencies by optimizing the use of community actors, yet  at  the  same  time  reach  very  large  numbers  of  poor  people.    Determining  whether  this  approach  is  worth  supporting requires rigorous evaluation to assess CDD’s effectiveness in various settings. And if it does work,  how  can  we  strengthen  its  ability  to  deliver  results  as  a  second  generation  of  CDD  programs  begins  to  emerge?  This study aims to explore these issues.    Communityâ€?  driven  development  (CDD)  is  an  approach  that  emphasizes  community  control  over  planning  decisions and investment resources.   Over the past decade, it has become a key operational strategy for many  national  governments—as  well  as  numerous  international  aid  agencies—for  the  delivery  of  services.    The  philosophy behind CDD is that involving communities in local development decisions is not only an inherent  citizen’s  right,  but  that  participation  can often  lead to a  better  use  of  resources geared toward meeting  the  needs of communities. CDD has become a popular development intervention because of its approach toward  empowering  local  decisionâ€?making  and  its  reputation  for  getting  resources  to  communities  efficiently.  The  approach  departs  from  traditional  approaches  to  development  by  enabling  communities  and  local  institutions—rather  than  central  governments—to  take  the  lead  in  identifying  and  managing  communityâ€? level  investments.    The  World  Bank  currently  supports  approximately  400  CDD  projects  in  94  countries  valued at almost $30 billion.  Over the past 10 years, CDD investments have represented between 5 and 10  percent of the overall World Bank lending portfolio.      There is a growing body of evidence documenting the impacts of CDD interventions. This paper reviews the  results  of  impact  evaluations  of  World  Bank  CDD  projects,  and  discusses  some  of  their  operational  and  research implications.    The  study’s  objectives  are  to  (a)  summarize  and  analyze  impact  evaluation  results  of  World  Bank  CDD  programs  over  the  past  25  years,  and  (b)  describe  critical  operational  and  research  implications  of  these  findings.    This  study  aims  to  generate  knowledge  in  answering  key  development  impact  questions  for  CDD  programs and by doing so, to improve the quality of operations using evidenceâ€?based approaches.    This study differs from earlier reports in several important respects.  First, the study team has updated the  data  sets,  focusing  especially  but  not  exclusively  on  the  latest  generation  of  CDD  impact  evaluations  from  2003 onwards.  Several of the evaluations were completed in the last five years and thus were not covered in  earlier  summary  reports.    Second,  whereas  some  recent  reports  (such  as  Mansuri  and  Rao,  forthcoming)  cover a very broad range of topics under the heading of “participatory development approachesâ€?—including  elections, decentralization, and schoolâ€?based management—it is worth fractionating out and understanding  more about a specific approach to participation from within the overall participatory approach universe, that  is World Bankâ€?supported CDD programs.  This study thus focuses much more narrowly on this specific subset  of CDD and its impact evaluation findings.  Third, as mentioned above, this paper attempts to focus more than  past studies on contextual factors and the operational implications of the findings.  What has worked under  what particular contexts, and why?     iv      The audience for this study is geared toward task teams and client/government project teams, particularly as  to  how these  findings  are  relevant  for  their  operational  work  and  the  global  participatory  research agenda  moving forward.    This paper looks at results across six key areas of interest:     1. What is the poverty or socioeconomic welfare impact of CDD/Social Fund programs?    2. Who benefits from these program interventions (poorest quintiles, women, ethnic groups)?  Do they  reach poor areas and poor households?     3. Do the programs improve access to and use of basic services?    4. Do  they  improve  social  capital  (using  the  standard  proxy  measurements  for  social  capital  –  trust,  collective action, association, groups and networks)?    5. Do  they  improve  local  governance  (participation  in  local  meetings,  satisfaction  and  increased  confidence with government officials, awareness of program activities, etc.)?    6. In conflictâ€?affected areas, do CDD operations have any impact on violent conflict?    The  study  examines  some  of  the  contextual  factors  behind  these  results  and  why  there  are  (or  are  not)  impacts  along  the  aforementioned  dimensions  in  order  to  derive  certain  operational  and  research  implications.        This review only looks at World Bank CDD programs with robust impact evaluations. Generally this means  that the evaluations use an experimental or quasiâ€?experimental design based on significantly large samples.  In other words, the studies have sufficiently large sample sizes as to claim some degree of internal validity,  are wellâ€?constructed, and have control groups.  Relevant information from accompanying qualitative studies,  project design documents, and completion reports were also used to strengthen the discussion. To enhance  quality control, the impact evaluations included in this study had to have undergone a peer review process.   Lastly, the study team spoke to several CDD experts, task team leaders, and evaluators about these programs  in order to understand more fully the contextual and design issues leading to the results.    Taking  these  criteria  into  account,  this  review  focuses  on  17  World  Bank  CDD  programs  in  South  and  East  Asia,  Africa,  Latin  America,  and  Central  Asia.    Many  of  these  programs  were  initiated  as  a  response  to  economic and financial crises, disaster, and conflict.  The results from the impact evaluations are as follows:    Impacts on socioeconomic welfare. Most of the 17 projects did not have an explicit objective of impacting  socioeconomic  welfare  during  that  phase  of  the  project.  Only  three  of  the  projects—those  supporting  primarily  livelihood  activities—had  explicit  goals  of  income  poverty  impacts.    Most  of  the  programs,  especially  Social  Funds,  focus  more  on  health  and  education  outcomes  and  access  to  services,  such  as  road  access.  In addition, income improvements were considered longâ€?term goals that were beyond the scope of  several  of  these  projects.  Nevertheless,  nine  projects  reviewed  in  this  study  reported  on  income  poverty  impacts  as  part  of  their  evaluations.    Out  of  the  nine,  seven  had  statistically  significant  positive  impacts  on  household living standards and welfare.      v    Poverty  targeting.    Overall,  most  programs  did  well  on  geographical  poverty  targeting  and  selecting  poor  areas to  operate.  Programs  that  made  use  of  poverty maps  and the  latest  national  statistical  data  were  the  most  effective  at  geographical  targeting.    At  the  household  level,  most  CDD  programs  invested  in  public  goods—such  as  roads,  bridges,  schools,  and  health  centers—that  benefited  the  broader  poor  and  nonâ€?poor  community.  However,  programs  were  generally  found  to  benefit  more  poor  than  nonâ€?poor  households  and  individuals.  Results are mixed based on certain socioeconomic characteristics (e.g. urban vs. rural) and types  of investments.      Access  and  utilization  of  services.  Evidence  across  the  programs  is  generally  strong  regarding  positive  impacts on access to and use of services, especially in health, education, and drinking water.    Social  capital  and  governance.    Only  a  few  of  the  impact  evaluations  measured  for  social  capital  (eight  projects)  and  local  governance  (five  projects),  so  the  evidence  is  scarce.    Based  on  the  limited  evidence  to  date, most projects have no impact on social capital, or at best mixed impacts. There is very little evidence of  social capital spillover effects.   For local governance, the evidence is more positive, however also mixed.     Conflict.  Five impact evaluations—for Afghanistan, the Philippines, Indonesia (two evaluations), and Sierra  Leone —examined CDD’s impact in conflict situations.  Not surprisingly, there is no impact on macro levels of  violence except in the case of insurgency groups in the Philippines.     In sum, this review found generally positive evidence  for poverty welfare reduction, poverty targeting, and  increased access to services. As most of the project objectives and financial resources focused on improving  access  and  service  delivery,  these  projects  achieved  their  stated  aims.    Evidence  is  limited  and  mixed  however, on governance, social capital spillovers, and conflict impacts.      This study also looked at issues related to cost effectiveness and rates of return, and generally found positive  results as compared to other modes of service delivery.  Many of these programs showed between 13 and 40  percent lower costs for equivalent works.      Based  on  project  and  evaluation  reports  and  discussions  with  project  stakeholders,  we  propose  several  hypotheses  to  explain  some  of  the  more  positive  impacts  on  poverty  reduction  and  improved  access  to  services. Some factors contributing to success in these areas include (a) establishing a more participatory and  inclusive  model  of  service  delivery,  which  allows  communities  to  identify  the  poorest  and  their  own  development needs;  (b) the  provision  of high  quality and  adequate  facilitation and  technical assistance;  (c)  capacity building for communities;  (d) utilization of poverty maps to target resources to poor areas; (e) block  grants of sufficient size over several years that are used for economically productive purposes; and (f) lastly,  flexibility in project design and implementation with an approach of “growth in learningâ€? over the medium  and longer term.    Project teams cite several issues leading to the mixed results in some areas.  Design issues include (a) lack of  clear  articulation  of  a  governance  reform  and  social  transformation  pathways;  (b)  the  “project  bubbleâ€?  phenomenon, in which broader reforms are difficult to enact; and (c) limitations to community mobilization  and  challenges  in  reaching  marginalized  groups.    Implementation  challenges  are  also  numerous,  including   financial disbursement delays; conflictâ€?affected situations; natural disasters; difficulties in coordination with  other  ministries;  operations  and  maintenance  concerns;  and  less  than  optimal  monitoring  and  evaluation  systems.  vi    This  study  points  toward  several  important  operational  and  research  implications  as  a  new  generation  of  CDD programs emerge:    Operational implications  ï‚· On the operations front, project designers should be clear what their project objectives are and what  can realistically be achieved, especially on the governance and social capital fronts within the social,  political, and economic contexts in which the projects operate.      ï‚· Teams  should  work  with  the  government  to  develop  clearer  roadmaps  for  institutional  and  governance  reform.  Institutional  change  will  take  time  and  there  is  no  straight,  quick  pathway  toward genuine reform.     ï‚· CDD  programs  are  beginning  to  have  longer  time  frames.    The  average  duration  of  CDD  projects  included  in  this  study  is  11.8  years,  implemented  over  several  phases.  These  longer  time  frames  provide an opportunity to discuss longerâ€?term institutional reforms and a roadmap for institutional  and social change.      ï‚· Project  teams  should  use  existing  poverty  maps  and  national  statistical  data  to  improve  their  geographical targeting toward poor areas. At the household level, several programs successfully used  community participatory targeting to reach the more marginalized groups.    ï‚· Project  designers  need  to  better  understand  the  impacts—both  positive  and  negative—of  development  interventions  in  general,  and  CDD  interventions  specifically,  in  fragile  and  conflictâ€? affected areas, and approach these situations cautiously in design.  Past experience shows that there  can be positive as well as negative impacts.    ï‚· Investing in capacity building and training for project staff and community groups is important for  longâ€?term sustainability, as well as potentially contributing to enhanced social capital formation and  improved local governance.    ï‚· Arrangements  for  operations  and  maintenance  (O&M)  of  subproject  investments  require  greater  attention  in  CDD  projects.    One  possibility  is  to  explore  piloting  various  options  for  O&M  arrangements within CDD programs.    ï‚· There is a need to experiment in developing stronger modes of collaboration with supplyâ€?side actors,  including  sectoral  lineâ€?department  actors  as  well  as  private  sector  agents  such  as  financial  intermediaries  and  banks  (for  microfinance  activities).    This  is  especially  true  for  service  delivery  effectiveness and sustainability.    Monitoring, reporting, and evaluation implications  ï‚· Monitoring,  reporting,  and  evaluation  systems  need  to  be  improved  in  order  to  provide  relevant  information  to  project  management  to  oversee  their  programs  and  adjust  course  as  needed.  Reporting  and  management  information  systems  in  particular  require  special  attention.    Rigorous  impact  evaluations  of  WB  CDD  programs  are  few  and  far  between;  only  17  or  so  rigorous  impact  evaluations have been conducted over the past 25 years.    vii      ï‚· Longitudinal studies are scarce. The impact evaluations included in this study measured on average  3.1  years  of  project  intervention,  despite  program  durations  averaging  almost  12  years.  There  is  a  tension  between  (a)  the  need  to  obtain  “quickâ€?  information  on  project  impacts  in  order  to  inform  future program directions, and (b) the time for some of these impacts to actually materialize. Certain  impacts by their nature may take longer to appear, such as governance spillover effects or education  learning outcomes.  Many more longâ€?term, high quality, and timely impact evaluations are required  to inform decision making.    Future Research Implications  On the research front, there are several areas worth pursuing:      ï‚· Examine over time social capital and local governance impacts to better understand clear pathways  of  reform.    Much  more  analytical  work  is  warranted  to  examine  the  pathway  for  longerâ€?term  local  governance  reform  and  the  enabling  factors  for  broadening  and  sustaining  the  impacts  of  communityâ€?driven development. Related to this, local leadership transformation is also important to  sustain these changes.    ï‚· Compare  CDD  approaches  versus  nonâ€?CDD  approaches  using  other  government  service  delivery  mechanisms.  None of the impact evaluations in this study compare “head onâ€? CDD approaches with  alternative  modes  of  government  service  delivery.  More  evidence  along  these  lines  would  greatly  inform policy decisionâ€?making.    ï‚· Compare why some programs are able to reach excluded and marginalized groups better than other  programs.  This  may  have  several  design  implications  in  terms  of  targeting  criteria,  training,  and  implementation modalities.    ï‚· Unpacking the black box of decision making is crucial. Very little is known or documented about how  decisions are made regarding the allocation of resources. The issue of elite and/or political capture is  also becoming a critical issue to explore further. How can we improve the accountability of decisionâ€? making  bodies  or  councils  in  terms  of  design  and  implementation?    Additional  qualitative  and  ethnographic work in this area would be valuable.      ï‚· Lastly,  issues  of  sustainability  came  up  repeatedly  in  this  review.  Project  reports  raised  two  dimensions  related  to  sustainability.  First,  CDD  programs  would  benefit  from  more  studies  on  technical  quality  and  operations  and  maintenance  to  sustain  physical  investments.  Second,  institutional  sustainability  is  important  as  well  as  linkages  with  government  agencies  and  private  sector  actors  to  sustain  investments.    As  mentioned  above,  longerâ€?term  evaluations  are  needed  to  assess whether program impacts attenuate or change.  viii    I.  Introduction    Communityâ€?driven  development  (CDD)  is  an  approach  that  emphasizes  community  control  over  planning  decisions and investment resources.   Over the past decade, it has become a key operational strategy for many  national  governments—as  well  as  numerous  international  aid  agencies—for  the  delivery  of  services.    The  philosophy behind CDD is that involving communities in local development decisions is not only an inherent  citizen’s  right,  but  that  participation  can  often  lead  to  a  better  use  of  resources  geared  toward  meeting  community  needs.    CDD  has  become  a  popular  development  intervention  because  of  its  approach  toward  empowering  local  decisionâ€?making and  its  reputation  for  getting  resources to  communities  efficiently.    The  approach  departs  from  traditional  approaches  to  development  by  enabling  communities  and  local  institutions—rather  than  central  governments—to  take  the  lead  in  identifying  and  managing  communityâ€? level investments.  The design of these programs has evolved considerably over the past two decades, and the  level of decisionâ€?making authority varies as well.  At their core, however, most of these programs generally  aim to improve the living conditions of poor communities through participatory means.     The World Bank currently supports approximately 400 CDD projects1 in 94 countries valued at almost $30  billion.    The  largest  number  of  projects  is  in  Africa,  followed  by  South  Asia  and  Latin  America.  However,  commitment amounts  are highest  in the  East  Asia  Pacific,  Africa, and  South  Asia regions.    Over  the  past  10  years,  CDD  investments  have  represented  between  5  and  10  percent  of  the  overall  World  Bank  lending  portfolio.      There is a growing body of evidence documenting the impacts of CDD/Social Fund (SF) interventions.2  This  paper focuses on the results of impact evaluations of World Bank CDD projects and discusses some of their  operational and research implications. This study is important for several reasons. First, impact evaluation is  a  priority  topic  among  the  500â€?memberâ€?strong  CDD  community  of  practice  at  the  World  Bank.  In  a  2010  survey, the practice community identified establishing sound monitoring and evaluation systems as the top  priority for future learning and training.  Second, this analysis hopes to strengthen the operational dialogue  among WB project task teams, government counterparts, and others by providing an evidence base regarding  various  types  of  CDD  approaches.  It  serves  as  a  reference  for  ongoing  and  future  CDD  operations  to  better  design  interventions  based  on  lessons  of  what  has  or  hasn't  worked  in  different  country  contexts.    World  Bank  task  teams  working  on  CDD  programs  are  the  primary  audience  for  this  paper,  but  the  lessons  raise  highly  relevant  issues  for  governments  and  development  practitioners  alike  who  are  interested  in  communityâ€?led poverty reduction efforts.                                                                                 1 This figure includes projects with CDD components. According to the WB CDD database, out of 400 active projects, 188 projects (47  percent) have a majority of their budget allocated for CDDâ€?related components.  2  For the purposes of this paper, Social Funds are included in the overall umbrella category of CDD programs.   1  II.  Objectives of the Study    The  study’s  objectives  are  (a)  to  summarize  and  analyze  impact  evaluation  results  of  World  Bank  CDD  programs over the past 25 years; and (b) to describe critical operational and research implications of these  findings.    This  study  aims  to  generate  knowledge  in  answering  key  development  impact  questions  for  CDD  programs and by doing so, improve the quality of operations using evidenceâ€?based approaches.    This study will examine the answers to the following six core questions:  1. What is the poverty or socio­economic welfare impact of CDD programs?    2. Who benefits from these program interventions (poorest quintiles, women, ethnic groups)?  Do they  reach poor areas and poor households?     3. Do the programs improve access to and use of basic services?    4. Do  they  improve  social  capital  (using  the  standard  proxy  measurements  for  social  capital  –  trust,  collective action, association, groups and networks)?    5. Do  they  improve  local  governance  (participation  in  local  meetings,  satisfaction  and  increased  confidence with government officials, awareness of program activities, etc.)?    6. In conflictâ€?affected areas, do CDD operations have any impact on violent conflict?    Some earlier summaries of impact evaluations did not go far enough in explaining why—or why not— there  are impacts along these six core dimensions. This study aims to fill this void by focusing on rigorous impact  evaluation  findings  as  well  as  some  of  the  possible  explanatory  factors  behind  those  impacts.    From  the  impact  evaluation  findings  and  a  review  of  the  project  documentation,  we  draw  out  several  operational  implications—such as design characteristics and implementation issues—affecting the results.  This is done  through  an  inâ€?depth  review  of  the  project  design  documents  (project  appraisal  documents  or  PADs),  complementary project studies, and Implementation Completion Reports (ICRs).  We also interview a subset  of  CDD  experts,  project  task  teams,  and  evaluators  to  hear  their  perspectives  on  what  worked  or  didn’t  in  their specific contexts and why.  Given our limited sample size, we are not able to propose a definitive set of  recommendations, but we can shed some light on several issues to consider in designing future studies and  operations.  Some of the issues we examine in relation to design features and country context are:    ï‚· What  are  some  of  the  common  country  political/economic  contexts—such  as  political  history,  economic context, institutional settings, and decentralization—prevalent in the CDD programs under  study?    ï‚· What are some of the common design features that work or fail and in what context?    ï‚· How do we maximize welfare outcomes in terms of block grant sizes and investments?    ï‚· What roles do institutional capacity building and training play in these operations?    2  ï‚· Why are we seeing limited impacts on social capital and governance in many cases? Is this an issue of  project design, intervention duration, country context, measurement, or some other factor?    ï‚· What areas need further research and data to inform operations?     Relationship and Complementarity with Other Studies  There  have  already  been  several  studies  over  the  past  few  years  summarizing  CDD/SF  evaluation  results  (Rawlings  et  al.  2003;  Wassenich  and  Whiteside  2003;  Mansuri  and  Rao  2004;    World  Bank  2007a).    Most  recently, Mansuri and Rao (forthcoming) conducted a much broader study of local participatory development  approaches worldwide.     This study differs from the previous studies in several important respects.  First, we have updated the data  sets, focusing especially but not exclusively on the latest generation of CDD/SF impact evaluations since 2003.   Several  of  the  evaluations  were  completed  in  the  last  five  years  and  were  not  covered  in  earlier  summary  reports.  Second,  whereas  Mansuri  and  Rao  (forthcoming)  cover  a  very  broad  range  of  topics—  including  elections, decentralization, and schoolâ€?based management—under the heading of participatory development  approaches, this study fractionates out a specific subset of participatory models for more inâ€?depth analysis,  e.g., World Bankâ€?supported CDD programs and impact evaluation findings. Third, one advantage of looking at  a specific slice of the participatory approach universe is that we are then able to focus more than past studies  on contextual factors and the operational implications of the findings.      The  study  is  geared  toward  task  teams  and  client/government  project  teams  and  how  these  findings  are  relevant for their operational work. It is meant to both complement and update previous research work in an  effort to provide operationally userâ€?friendly guidance.              3  III.  Study Methodology    Defining the Universe  Types of Projects Covered in this Study   There are several pitfalls in undertaking this exercise. First, there is the definition of what constitutes a CDD  program.  The CDD community generally looks at a broad typology of programs and variants, ranging from  community  control,  decision  making,  and  management  of  grant  funds—such  as  the  Indonesia  Kecamatan  Development Program (KDP), or the Philippines KALAHIâ€?CIDSS—to community inputs into the planning and  decisionâ€?making process of local governments through semiâ€?autonomous government bodies—for example,  some earlier Social Fund models in Bolivia and Nicaragua. The earlier Social Funds relied on oneâ€?off project  proposals  submitted  by  a  variety  of  local  actors,  including  NGOs  and  community  groups.  There  are  also  numerous projects with CDD components, but they account for a fraction of the total project budget. For the  purposes of this study, we include programs that include community control and management of funds and  emphasize  community  inputs  into  the  planning  and  decisionâ€?making  process.  These  CDD  and  SF  programs  are included in the World Bank’s CDD database.    Second,  this  review  focuses  only  on  impact  evaluations  for  Worldâ€?Bankâ€?supported  projects.  There  are  hundreds of community projects (usually much smaller in scale) funded by various NGOs, governments, and  donors  all  over  the  world.    This  study  limits  itself  to  Worldâ€?Bankâ€?supported  projects  in  order  to  keep  the  analysis manageable and focused.  In the future, it may be worthwhile to include nonâ€?World Bank community  projects in the data set.    This study also focuses on those WB CDD projects that are multisectoral in nature and do not include sectorâ€? specific  programs  such  as  schoolâ€?based  management,  community  healthcare  management,  or  natural  resource  management  interventions.    Again,  there  are  hundreds  of  these  programs  supported  by  national  governments  and  other  international  donors.    These  types  of  programs  are  not  included  in  this  study  for  several  reasons,  including  time  and  human  and  financial  resource  constraints.  Furthermore,  several  other  recent  education  and  health  metaâ€?analyses  cover  these  topics  in  a  much  more  comprehensive  and  holistic  manner than could be described here (Bruns et al. 2011; Barreraâ€?Osorio et al. 2009).      Types of Evaluations Covered  The  literature  review  drew  from  the  Development  Impact  Evaluation  Initiatives  (DIME)  database,  which  is  the  official  impact  evaluation  database  of  the  World  Bank  and  from  the  World  Bank  CDD  community  of  practice.  In order to keep the study manageable and report rigorous findings, the study team stayed within  several parameters:      i. Impact evaluations had to use an experimental (random assignment) or quasi­experimental design based  on  sufficiently  large  samples.  In  other  words,  the  studies  have  sufficiently  large  sample  sizes  to  claim  some  degree  of  internal  validity,  are  wellâ€?constructed,  and  have  a  control  group.    Impact  evaluations  aim  to  establish  whether a particular intervention changes outcomes for the beneficiary population.  The central issue for all impact  evaluations  is  determining  the  counterfactual:  What  would  have  happened  to  the  beneficiaries  had  they  not  received the intervention?  Since we cannot observe the counterfactual, control or comparison groups are used as a  proxy for the state of beneficiaries in the absence of the intervention.  Using an experimental or quasiâ€?experimental  design allows us to determine project attribution.  4  Four  of  the  17  impact  evaluations  described  in  this  review—Afghanistan  NSP,  Indonesia  PNPMâ€?Generasi,  Nepal  Poverty  Alleviation  Fund,  and  Sierra  Leone  GoBiFo—use  an  experimental  (random  assignment)  design,  while  the  rest  are  primarily  quasiâ€?experimental  designs  using  propensity  score  matching  or  matched comparisons. An experimental design is considered the most methodologically rigorous for impact  evaluations  but  may  not  be  possible  due  to  political,  logistical,  or  other  important  considerations.  This  study points out the methodology being used under each evaluation description in section VI and Annex A.       This  study  does  not  report  on  projects  that  used  only  qualitative  studies  to  measure  impacts,  since  by  definition,  those  study  findings  cannot  be  generalized  for  the  project.    This  is  not  to  say  that  qualitative  studies  are  not  important. However,  many  smallâ€?scale  studies  lack  the  sample  size  necessary  to generate  statistically  significant  results.    To  the  extent  that  qualitative  studies  inform  findings  from  a  quantitative  study and provide possible explanations and a more comprehensive or nuanced understanding of program  impacts, we report on some of those findings. Indeed, some of the best impact evaluations in our study use  a mixed methodology of quantitative and qualitative techniques.    ii. Peer–Reviewed. As an added level of quality control, we only include impact  evaluations that have been  peerâ€?reviewed.    This  is  to  ensure  that  the  study  methodology  has  been  vetted  and  reviewed  for  appropriateness and the findings have been technically reviewed.  Several evaluations were excluded from  this study because they had not been peerâ€?reviewed and finalized.    The abovementioned criteria narrowed the field down to 17 WB CDD programs for our review (see Table 1).   For each of these CDD programs, we reviewed the impact evaluations as well as complementary studies and  projectâ€?related  documents  such  as  project  appraisal  documents  (PADs)  and  implementation  completion  reports (ICRs). The team also contacted several CDD experts and some of the project and evaluation teams to  understand from their perspective what worked or failed in their specific contexts.      Two important qualifications should be noted by the reader.  While this study focuses upon 17 WBâ€?supported  CDD  programs,  these  findings  should  not  be  interpreted  as  representative  of  the  400  or  so  ongoing  CDD  programs in the Bank’s portfolio.  If anything, the study focuses on those CDD programs with rigorous, high  quality impact evaluations in place. If these programs were able to implement these evaluations, it is likely  this subset may represent some of the better programs in the portfolio. There is no mandatory requirement  at  the  WB  to  conduct  an  impact  evaluation,  although  it  is  regarded  as  best  practice  for  strategic  and  innovative projects.  Having an impact evaluation in place is primarily at the initiative and discretion of the  World Bank task team and the respective government.  It is important to keep this in mind in reviewing the  evidence.    A second important qualification is that the selection of these 17 programs with impact evaluations does not  mean  that  there  are  no  important  lessons  to  be  drawn  from  other  CDD  programs  in  the  portfolio.    Many  programs  provide  valuable  insights  and  lessons  learned  as  a  critical  part  of  their  adaptive  learning  and  management approach.  5  Table 1. World Bank CDD Programs Covered in this Study    Name of CDD Program  Total Project Period  Total Amount for the  Types of Subâ€?projects Funded under IE Phase  (Program Period  IE Phase *  Relevant for IE)  (WB amount)  $ million  SOUTH ASIA  1  Afghanistan: National Solidarity Program II  2003 to 2015  $120 M  Infra (water, sanitation, rural road, irrigation, etc), and  (2006 to 2011)  (WB $120 M) with   human capital development.    additional financing of  $75 M  2  India:  Andhra Pradesh District Poverty Initiative  2000 to 2016  $ 118.2 M  Microfinance activities: household dairy (30%),  Program (DPIP)    (2000 to 2006)  (WB $111 M)  agriculture (29%), nonâ€?farm trade and sheep rearing  (10%)  3  Nepal: Poverty Alleviation Fund II  2004 to 2014  $109 M   Income generation (livestock, agriculture etc) (72%),  (2008 to 2014)  (WB $100 M) with   smallâ€?scale infra and services (micro irrigation, water  additional financing of   supply etc) (17%).  $65 M      EAST ASIA        4  Indonesia:  Kecamatan Development Program II  1998 to 2014  $421.5 M   Rural village infra (70%) and revolving funds and social  (KDP2)  (2002 to 2007)  (WB $111.3 M)  services (30%).  5  Indonesia: PNPMâ€?Rural Generasi Sehat dan Cerdas  1998 to 2014  $98 M  Education activities (56%),  health activities (44%) to  (PNPM Generasi)  (2007 to 2010)  meet 12 education and maternal and child healthcare  targets  6  Indonesia: KDP/BRA  2006 to 2007  $21.7 M   Economic activities such as livestock (89%) and rural  (same)  infra (10%).  7  Indonesia: Urban Poverty Program II (UPP2)  1999 to 2014  $126.9 M  Revolving funds for providing microâ€?credit services,  (2002 to 2008)   (WB $100 M)     small infra improvements (roads, drains, water,    sanitation, etc).  8  Philippines:  Kapitâ€?bisig Laban Sa Kahirapan –  2002 to 2014  $182 M   Rural infra (water systems, school buildings) (50%);  Comprehensive and Integrated Delivery of Social  (2003 to 2010)  (WB $100 M)  basic transport infra (28%); and community enterprise  Services Program (KALAHI CIDSS)     facilities.               6  Total Amount for the  Total Project Period  IE Phase *    Name of CDD Program   (Program Period  Types of Subâ€?projects Funded under IE Phase  (WB amount)  Relevant for IE)  $ million    AFRICA  9  Senegal: Programme National d’Infrastructures  2000 to 2012  $42.9 M   Water supply, schools, health centers and health  Rurales (PNIR)  (PNIR/PLDP)  (WB $28.5 M)  posts, livestock projects, promotion of women’s  (2000 to 2005)  activities, agriculture, youth and sports, commerce  and miscellaneous.  10  Sierra Leonne: GoBifo  2005 to 2010  $2.5  M  (JSDF,  Italian  Local public goods construction, agricultural  (2005 to 2009)  Govt)  production and livestock management, skills training  and incomeâ€?generating activities, and social projects  such as youth sport clubs.  11  Tanzania Social Action Fund Project II **  2000 to 2013  $ 178.5 M   Education, health, water, roads, irrigation, markets,  (2004 to 2013)  (WB  $150  M)  with  an  and support to vulnerable groups.    additional  financing  of   $35 M   12  Zambia Social Recovery Project II  1991 to 2005  $47.7 M   Education (80%), health (12%) and nutrition and  (1995 to 2000)  (WB $30 M)  economic infra      LATIN AMERICA AND CARRIBEAN        13  Bolivia Social Investment Fund II  1987 to 2004  $70 M  Water, school construction, health and sanitation.   (1993 to 1998)   (WB $40 M)  14  Honduras Social Investment Fund III  1990 to 2012  $117.9 M  Smallâ€?scale infra, furniture and equipment, and  (1995 to 1999)   (WB $27.4 M)  training for education, health, water and sanitation.   Also child/elderly care centers and orphanages.  15  Nicaragua Emergency Social Investment Fund II  1992 to 2006  $172 M   Social infra (70%), education, health, economic infra,  (1994 to 1997)  (WB $26.5 M)  social services and environment  16  Peru Social Fund (FONCODES)   1993 to 2000  $430 M   Social infra (school, water and sanitation), Social  (1993 to 1997)  (WB $150 M)  assistance, economic infra (rural roads, bridges,  irrigation system) and productive projects.     EASTERN AND CENTRAL ASIA        17  Armenia Social Investment Fund   1996 to 2012  $20 M   School rehabilitation, potable water, irrigation and  (1996 to 2000)  (WB $12 M)  health facilities   *Total costs including WB and counterpart contributions  **  Tanzania SF relates to targeting study only      7  IV.  Conceptual Framework behind CDD Interventions   Figure 1. Illustrative CDD Results Chain   Note: Not all CDD programs aim for all these results. Examples above are illustrative only.  Objectives vary by project.  It is important to understand the theory behind CDD interventions and how they operate.  As discussed in the  following section, CDD programs historically grew out of situations of crises (financial, disaster, conflict) or  when  governments—usually  newly  installed  administrations— sought  a different mode of service  delivery.  Existing  systems  were  considered  ineffective  in  engaging  citizens  or  delivering  services  in  an  effective  manner.      CDD program designs usually involve communities undertaking a villageâ€?level participatory planning process  3 with the help of project facilitators or local government officials.  Communities identify their priority needs  and the plans or proposals are sent to an interâ€?village forum of village representatives or higher level forums  at the district or provincial levels for decision making. These decisionâ€?making forums vary widely across the  CDD portfolio. There is a large variation in design across different programs, and even between earlier and  later  phases  of  the  same  project.    Some  programs  consist  of  decisionâ€?making  bodies  completely  elected  by  communities (Indonesia, Philippines, and Afghanistan); in other projects, higher level government bodies or                                                                     3 The earlier design of many social funds did not necessarily derive from a formal process of local planning or community needs assessments. NGOs, local governments, or community groups could submit proposals. 8  the  Social  Fund  semiautonomous  executing  agency  play  a  greater  role  in  deciding  on  the  selection  of  proposals  (earlier  SFs  in  Bolivia,  Honduras,    Nicaragua,  and  Zambia,  for  example).  Technical  ministries  normally  provide  input  into  the  community  plans  to  ensure  they  do  not  duplicate  ministerial  technical  programs and are consistent with ministerial plans for that area.  For example, if a community wishes to build  a school, the local education department ensures that they will be able to place teachers in that school and  that the school is in line with sectoral plans.     Unlike  many  development  projects,  CDD  programs  do  not  predetermine  the  nature  of  the  interventions.   Some programs have a range of investments across sectors; others use an “open menuâ€? with a negative list.   The multisectoral nature of many of these CDD programs can be seen as one of its strengths, since the flexible  use of investments can help meet selfâ€?identified community needs. They are not just limited to one sector or  the requirements of one sectoral ministry.  If the plans are funded, communities manage and implement the  activities or contractors are hired for construction works.  The project normally trains and provides technical  assistance  to  communities  to  plan,  implement,  and  manage  these  activities.  Investments  are  meant  to  be  community  demandâ€?driven.  The  design,  structure,  and  resources  allocated  to  each  stage  of  the  process  can  vary greatly.    An illustrative results chain for CDD programs is outlined in figure 1.  Communities or local institutions take  the  lead  in  identifying  and  managing  communityâ€?level  investments,  be  they  rural  infrastructure,  social  services, or incomeâ€?generating activities.  Not all CDD programs share the same objectives; for example, some  focus more on poverty reduction and service delivery than community empowerment or local governance.       However,  what  is  equally  important  in  looking  at  CDD  project  design  are  the  many  assumptions/risks  involved  in  this  approach.    There  are  risks  that  the  technical  assistance  and  capacity  building  may  not  be  sufficient  at  the  local  level  to  facilitate  community  involvement  and  effective  management  of  resources.  Or  that decision making can  be  done  in  a participatory manner  that allows for community  representation and  voice rather than elite capture or further reinforcement of existing patronage systems.  All these design risks  must  be  taken  into  account  when designing  and  implementing a  CDD  program. The  achievement  of project  objectives  and  real  reform  will  not  happen  without  numerous  factors  being  in  alignment,  including  many  areas that are outside the control of communities and the project itself.     Longer Term Vision of CDD Programs  Debates about CDD often lead to discussions regarding what are their ultimate objectives.  Are they merely a  transitional instrument for service delivery in times of crisis and fragile situations, or do they have a longerâ€? term  role  in  decentralized  planning  and  institutional  frameworks?  Related  to  these  issues  are  questions  regarding  their  relationship  to  local  government  entities.  CDD  critics  often  argue  that  CDD  programs  take  away  precious  resources  that  should  otherwise  be  channeled  directly  through  local  government  bodies  to  build their capacities and allow them to deliver services.  By channeling funds directly to communities, these  programs are competing with local government actors and creating alternative power centers.      CDD  proponents  counter  that  in  countries  where  there  is  no  locally  elected  form  of  government  or  an  extremely  weak  state  at  the  local  level—such  as  in  fragile  and  postâ€?conflict  situations—it  is  simply  not  possible to channel funds directly to local government entities, especially if there is an urgency in delivering  services or in times of postâ€?crisis.   In the case of  some more stable middleâ€?income countries, however, the  rationale may be quite different.  Proponents argue that citizens have already had decades of experience with  9  governments  not  delivering  quality  services  at  the  local  level  and  thus  a  change  of  approach  –  from  the  “bottom upâ€? ­ is needed.  Putting planning and resources in the hands of community groups allows for moreâ€? tailored  local  responses  and  availability  of  services.  Some  believe  that  the  CDD  approach  also  empowers  communities to demand better governance and services from local government in the future.  They further  argue  that  there  is  no  evidence  to  indicate  that  supporting  and  empowering  local  communities  hurt  local  governments.    These  healthy  debates  will  continue.    However,  what  we  are  seeing  in  terms  of  CDD’s  evolution  is  in  many  ways a merger or hybrid of the aforementioned issues.  Many programs, especially the ones developed in the  last  15  years,  channel  funds  through  established  government  ministries  and  provide  annual  block  grant  allocations.  Almost all the programs also contain activities to build the capacity of government actors at the  national  and/or  local  levels  to  undertake  participatory  or  area  planning,  and  improve  their  reporting,  monitoring,  and  technical  service  delivery  skills.    In  several  cases—Indonesia,  Senegal,  the  Philippines,  and  Afghanistan,  for  example—the  CDD  programs  provide  support  to  governments  in  thinking  through  their  governance and decentralization reform agendas and procedures.       In conclusion, CDD designs vary greatly according to the local country context and history. They have evolved  significantly over the past 25 years.  The latest CDD program designs maintain the principle of communityâ€?led  development and empowerment, and work closely with government agencies in achieving their development  goals.  As  we  will  discuss  in  later  chapters,  however,  there  is  a  critical  need  in  many  situations  to  think  through more carefully these pathways of governance and social reform.                  10  V.  Country Contexts     The  country  context  is  extremely  important  when  trying  to  understand  the  existence  and  impact  of  these  programs.  For example, countries emerging from decades of authoritarian rule, financial crises, and/or civil  strife pose enormous challenges for undertaking effective development programs, not only communityâ€?based  endeavors but many types of effective service delivery. The first social fund was introduced in Bolivia in 1987  to  provide  temporary  employment  and  stabilization  of  social  services  during  a  time  of  economic  crisis  and  adjustment. The emphasis was clearly on delivering education and health services as a response to a major  structural  adjustment  program  when  social  expenditure  spending  was  being  cut.      Countries  facing  severe  economic  crises  and/or  adjustment  programs  followed  suit,  especially  in  Latin  America  (Honduras,  Nicaragua, Peru, etc.) and Africa (Zambia). These were the first generation of social funds created to serve as  shortâ€?term  safety  nets  to  soften  the  impact  of  structural  adjustment  policies  on  the  poor  and  provide  emergency  relief  by  creating  temporary  employment. They  were  usually  managed  by  a  semiâ€?autonomous  government body that appraised and supervised the implementation of subprojects proposed by community  groups, NGOs, or local governments.  These initial programs were assumed to have a short life span.    The emphasis on local participation and enabling communities and local institutions to take the lead, rather  than central governments, continued with the design of CDD programs in the later 1990s and 2000s.  In many  countries,  public  investments  were  determined  by  central  ministries  in  a  “topâ€?downâ€?  manner.  Project  development  objectives  emphasized  improving  local  governance—for  example,  Indonesia’s  KDP  following  the 1998 Asian economic crisis and severe drought, or Afghanistan’s National Solidarity Program in 2003—as  well as decentralized delivery of services  and community empowerment, such as the Kalahiâ€?CIDSS project in  the  Philippines,    India’s  Andhra  Pradesh  District  program,  or  Nepal’s  Poverty  Alleviation  Fund.        In  many  cases,  governments  faced  the  challenge  of  establishing  their  credibility  in  delivering  results  on  the  ground.   The CDD model is not monolithic and varies enormously, as one would expect depending on certain country  contexts, government champions, and WB task teams.  Several communityâ€?driven programs—in Sierra Leone,  Nepal, Afghanistan, or Aceh in Indonesia—were initiated to respond to other types of crises such as conflicts  and natural disasters.  The following table illustrates the main typology of country contexts in which CDD interventions have been  launched.  There  are  three  general  contexts:  (1)  a  financial/economic  crisis,  where  CDD  was  seen  as  an  emergency response to delivering services quickly to communities; (2) fragile and conflict situations where  “nothing  else  works;â€?  and  (3)  middleâ€?income  countries  that  turn  to  CDD  approaches  to  tackle  stubborn  pockets  of  rural  poverty  while  directly  engaging  communities  to  participate  in  development  planning,  decision making, and management.          11  Table 2.  Country Contexts and Origins of the Programs  Middleâ€?income situations of  High Fragility or Postâ€?Conflict  Financial Crisis/Structural Adjustment  entrenched or worsening  Situations  poverty   Bolivia, 1987   Afghanistan, 2003   Indonesia KDP/UPP, 1998   Honduras, 1990   Nepal, 2004   Philippines, 2002   Zambia, 1991   Sierra Leone, 2005   India AP, 2000   Nicaragua, 1992   Indonesia  KDPâ€?BRA,  2006   Peru (w/conflict), 1991  (postâ€?tsunami, postâ€?conflict)     Armenia  (transition  from  planned  economy, earthquake), 1996     Indonesia  KDP/UPP  (East  Asian  financial crisis, w/ drought), 1998  Source: Author compilation from PADs    The following country examples illustrate the wide variety of country contexts and circumstances in which  4 these CDD and Social Fund programs were born.         ï‚· Bolivia. Bolivia is considered the birthplace of social funds.  The first Emergency Social Fund began  in 1987 as a response to the financial crisis. It was an emergency measure to cushion the impact of  the government’s stabilization policies in the midâ€?1980s by generating employment rapidly for those  displaced  by  the  economic  crisis.    The  country  suffered  from  weak  institutional  capacity  characterized  by  centralization  in  decision  making;  weak  policy  making,  planning,  and  implementation; inefficiencies in resource allocation; and inadequate financial management.  A 1990  integrated household survey showed that the incidence of rural and urban poverty was 55 percent  and higher. The education level for most Bolivians was limited, and health indicators were the worst  in the continent.      ï‚· Peru.  In July 1990, the situation was characterized by hyperinflation of 400 percent, deep recession,  high unemployment and severe reduction of family income, with real wages reduced by oneâ€?half in  three years.  Poverty was widespread; 53 percent of the population fell below the poverty line, with  21 percent of the population living in extreme poor.  The situation was characterized by high infant  and maternal mortality rates, malnutrition in some rural areas as high as 73 percent, high prevalence  of  endemic  diseases,  and  new  epidemics  such  as  cholera.    In  education,  29  percent  of  the  rural  population was illiterate and 44 percent of rural women could not read or write.  Large areas of Peru  were still considered insecure due to the continuation of terrorist activities.  The government of Peru  created FONCODES  (Fondo Nacional  de  Compensacion Desarrollo  Social,  or  National  Fund for  Social  Compensation and Development) in August 1991 in an effort to mitigate the social costs of economic  adjustment  through  financing  of  smallâ€?scale,  shortâ€?term  social  investment  projects  of  immediate  impact on the poor and most vulnerable.    ï‚· Sierra Leone. More than a decade of armed conflict in Sierra Leone ended in January 2002. A 2004  household survey showed that 70 percent of people lived below the poverty line of $1 per day, with                                                                     4 The information in this section comes from a review of program appraisal documents, completion reports, and other country reports  for the time period of the program.  12  26 percent living in extreme poverty.  Life expectancy was 42 years, the second lowest in the world,  and the adult illiteracy rate was 79 percent.   The CDD project, GoBiFo (which means “Move Forwardâ€?  in the national lingua franca, Krio) began in 2005.      ï‚· Nepal.   When the Nepal Poverty Alleviation Fund began in 2004, 40 percent of the population lived  below the poverty line. Ethnic minorities and lower caste communities in remote areas, and women,  especially female heads of households, lagged seriously behind in terms of incomes, assets, and other  human  development  indicators.    The  Maoist  insurgency,  which  began  in  1996,  was  growing  in  intensity,  fuelled  by  the  country’s  deep  inequalities,  poor  governance,  and  corruption.  The  government  knew  that  more  direct  and  quick  interventions  were  required  to  help  marginalized  people, as well as send a clear signal that the exclusionary practices of the past were unacceptable.   The government wanted to put in place a program to address poverty and social exclusion, including  in conflictâ€?affected settings.  The government, with WB support, began the Nepal Poverty Alleviation  Fund to address these challenges.    ï‚· India, State of Andhra Pradesh. For decades, the main rural sector issue throughout India, including  Andhra Pradesh, was the lack of success of public antiâ€?poverty programs to reduce poverty.  Previous  government  schemes  had  several  deficiencies:    (a)  low  beneficiary  involvement  in  planning,  implementation and monitoring; (b) supplyâ€?driven implementation, primarily by numerical targets;  (c)  poor  targeting;  and  (d)  inefficient  management.    In  Andhra  Pradesh  in  2000,  30  percent  of  the  population lived below the poverty line; malnutrition among children aged 0–6 years was about 30  percent,  and  the  female  literacy  rate  (33  percent)  was  one  of  the  lowest  in  India.    The  Andhra  Pradesh District Poverty Initiatives Project aimed to develop a demandâ€?driven approach with a focus  on community ownership, in which the poor themselves would identify investments and participate  in their implementation.    ï‚· Armenia.    The  country  was  in  a  poor  economic  state  in  1996.  Fiftyâ€?five  percent  of  the  population  lived in poverty, with many households moving in and out of poverty. Economic hardships stemmed  from the transition from a centrally planned economy, a devastating earthquake, and war and civil  disorder in the bordering states. In response to these difficulties, the government of Armenia created  the Armenian Social Investment Fund as an autonomous and demandâ€?driven agency to finance smallâ€? scale projects for the rehabilitation of basic social and economic infrastructure.    ï‚· Indonesia. When the Kecamatan Development Program (KDP) began in 1998, Indonesia was in the  midst of an extreme financial and political crisis. A prolonged drought had pushed the rural poor in  many  regions  deeper  into  poverty.  Economic  growth  declined  to  minus  14  percent  in  1998,  in  marked contrast to the 8 percent annual growth rates of earlier years. While the country had made  significant progress for decades on poverty reduction, those gains were quickly reversed as millions  of  rural  poor  fell  below  the  poverty  line  and  significant  numbers  of  people  increased  their  vulnerability  and  clustered  around  the  poverty  threshold.  The  economic  crisis  brought  to  the  forefront  the  vulnerability  of  Indonesia’s  approach  to  economic  development  and  governance.  Thirtyâ€?two years of authoritarian rule had stifled localâ€?level initiative and decision making, as well as  local community organizing. Government programs in general had not been responsive to local needs  and  the  mismanagement  of  public  projects  and  funds  had  led  to  general  disillusionment  with  the  government’s  ability  to  provide  essential  services.    KDP  began  in  this  context  and  was  designed  to  13  improve local governance, increase the participation of communities in their own development, and  provide increased economic opportunities at the village level in the poorest areas.    The  commonality  among  these  highly  diverse  country  examples  was  the  feeling  that  existing  systems  were  not working or could not be relied on to deliver services in a timely manner and improve governance.  Many  of the countries also suffered from overly centralized systems. CDD programs were an attempt to reverse the  traditional  “top  â€?downâ€?  way  of  delivering  services  by  letting  communities  take  the  lead  in  identifying  and  implementing smallâ€?scale investments in a number of sectors.  As governments and the wider development  community continue to face challenges regarding how to reach poor people in situations of fragility and postâ€? conflict, economic crises, or in countries where the government bureaucracy and local administrations have  not been able to meet the needs of the poor and excluded, CDD is but one approach to address some of these  problems.    The  next  few  sections  discuss  what  has  been  learned  about  CDD  interventions  and  their  effectiveness in different settings.    Project Descriptions  Table 1 provides a quick synopsis of the various project designs, their time duration, budget size, and what  types of activities were funded.  Annexes A and B provide more detailed discussion of project development  objectives, background, and details for each project.    As  the  table  shows,  the  earlier  Social  Funds  in  Latin  America  and  Zambia  supported  primarily  education,  health,  and  water  and  sanitation  activities  in  keeping  with  the  original  intention  to  act  as  “safety  netsâ€?  in  response to cutbacks in social spending due to the economic crises and structural adjustment programs.  The  other  CDD  programs  funded  a  variety  of  subproject  activities  in  smallâ€?scale  economically  productive  infrastructure such as rural roads, bridges, irrigation, and markets; some social infrastructure for schools and  health centers; and microâ€?finance or incomeâ€?generating activities.      14  VI.  Impact Evaluation Results and Findings     We now turn to the results of the impact evaluations.  Projects are categorized by region, with detailed data  on projects and their specific evaluations in Annexes A and B.  When introducing a specific project, we give a  brief description of the project followed by the type of impact evaluation methodology used and the impact  evaluation  results.    Table  3  at  the  end  of  this  section  provides  a  synopsis  of  the  impact  evaluation  results  along the six main areas of inquiry.    1.  Were the Goods Delivered?   In examining possible impacts, we must first determine if the project proceeded as planned and whether the  services and outputs were actually delivered.  This will provide some indication whether the lack of impact is  due to poor or weak project implementation, or is a question of poor project design.  The project completion  reports  (ICRs)  indicate  that  for  all  17  projects,  the  achievement  of  development  outcomes  and  project  implementation  at  the  time  of  the  evaluation  was  satisfactory.    All  projects  listed  social  services  and  subprojects  delivered,  or  incomeâ€?generating  activities  in  place as  the  case  may  be.    Given  that services  and  “outputsâ€? are indeed what governments are supposed to deliver, and the difficult environments many of the  projects were operating in, these accomplishments are noteworthy in and of themselves.  For the most part,  social  funds  concentrated  on  education  and  health  activities,  while  other  CDD  programs—such  as  in  Indonesia,  the  Philippines,  India,  and  Sierra  Leone—concentrated  primarily  on  rural  smallâ€?scale  infrastructure and/or incomeâ€?generating activities.    All projects did encounter implementation difficulties however as will be described in more detail in section  VIII.    Problems  included  currency  devaluation  and  fluctuation,  delayed  financial  disbursements,  natural  disasters,  violent  conflict,  changes  in  project  personnel,  and  difficulties  recruiting  and  retaining  qualified  project  staff.    Based  on  the  final  reports,  it  seems  that  projects  were  able  to  cope  with  and  manage  these  problems and risks, at least to the point where outcomes and implementation were ranked as satisfactory.    2.  Poverty Reduction Impacts  Many  CDD/SF  projects,  not  all,  look  at  impacts  on  household  economic  welfare,  measured  in  terms  of  consumption, expenditure, assets, and/or income.  Of the 17 projects reviewed in this study, only four of the  projects—India’s  Andhra  Pradesh  DPIP,  Nepal’s  Poverty  Alleviation  Fund,  Indonesia’s  KDPâ€?BRA,  and  the  Zambia  Social  Investment  Fund—contained  within  their  project  development  objectives  the  explicit  aim  of  “improving  opportunities  for  the  poor  and  vulnerable,  especially  women,  to  meet  their  own  social  and  economic development objectives;â€? “reducing rural poverty;â€? and “improving living conditionâ€? (see annex A).  Most  of  these  projects—more  than  the  other  CDD  projects  in  this  study—focused  on  livelihood  activities.   Given that poverty reduction and welfare changes are generally longer term goals, this is not surprising for  project  objectives  over  a  threeâ€?  to  fourâ€?year  phasing  period.    Many  projects,  especially  social  funds,  focus  primarily  on  improving  access  and  utilization  of  services—such  as  school  enrollment  and  health  center  access—rather than direct welfare gains. Their project objectives focus on improving coverage and quality of  health,  education,  and  other  social  services  rather  than  income  gains.  Nevertheless,  despite  it  not  being  an  explicit  goal  of  the  projects,  many  of  the  impact  evaluations  did  measure  household  welfare  in  terms  of  consumption and/or assets in order to examine poverty targeting and possible spillover effects on household  welfare.   Nine impact evaluations reported on household welfare and poverty impacts.  Seven CDD programs  15  showed  positive  impacts,  while  only  two—Afghanistan  and  Indonesia  Urban  Poverty  Program  (UPP2)— showed  no  poverty  impacts.      For  Afghanistan,  the  lack  of  welfare  impacts  for  the  Afghanistan  NSP  is  not  surprising  since  the  midline  survey  report  clearly  stated  that  the  evaluation  was  conducted  prior  to  the  completion  of  many  projects  and  focused  primarily  on  creation  of  local  councils  and  the  socialization  and  planning process.    SOUTH ASIA  Afghanistan  National  Solidarity  Program  II  (NSP2).  This  program  began  in  2003  after  the  change  in  government.  The objective of the program was to lay the foundations for a strengthening of communityâ€?level  governance  and  to  support  communityâ€?managed  subprojects  comprising  reconstruction  and  development  that improve access of rural communities to social and productive infrastructure and services.  Subprojects  primarily  consisted  of  rural  infrastructure  such  as  water  and  sanitation,  transportation,  power,  and  education.     The  midline  survey  for  Afghanistan  NSP  was  conducted  in  2009,  measuring  after  two  years  of  implementation.  (Beath et al. 2010)    Followâ€?up surveys are planned for 2011 and 2013–15.  The evaluation  follows a  randomized  design  across  500  villages  in  ten  districts.    The  estimates represent  the  intermediate  impacts  of  the  program  when  community  development  councils  (CDCs)  have  been  formed  and  are  operational, but prior to the completion of many subprojects funded by NSP.  The midline results show that  there  are  improvements  with  regard  to  villagers’  perceptions  of  wellâ€?being.    Villagers  are  more  optimistic  about future economic changes and report that their economic situation has improved.  However, there are  no impacts on objective measures of economic welfare, such as levels of household income or consumption,  which may not be all that surprising given that subprojects were not yet fully completed during the interim  round and the objective of the program at this stage focused more on local governance and how the creation  of CDCs and the selection of development projects affected political and social attitudes and outcomes.      India Andhra Pradesh District Poverty Initiative Program (DPIP). In India’s state of Andhra Pradesh, many  public antiâ€?poverty programs were failing to reduce poverty due to the lack of community involvement and  inefficient program management. Begun in 2000, Andhra Pradesh’s DPIP aimed to improve opportunities for  the rural poor to meet priority social and economic needs in the six poorest districts of Andhra Pradesh. To  achieve this, the project would (a) help create selfâ€?managed grassâ€?roots institutions; (b) build the capacity of  established  local  institutions,  especially  the  Gram  Sabha/Panchayats  and  the  government  of  Andhra  Pradesh’s line departments, to operate in a more inclusive manner in addressing the needs of the poor; (c)  support investments in subprojects proposed by grassâ€?roots institutions of the poor to accelerate their entry  and expand their involvement in social and economic activities; and (d) improve access to education for girls  to  reduce  the  incidence  of  child  labor  among  the  poor.  The  focus  was  on  building  selfâ€?help  groups  and  supporting incomeâ€?generating activities.      The impact evaluation consisted of household panel surveys conducted in 2004 and 2006 to look at poverty  effects after 2.5 years of program exposure.  The evaluation found significant economic gains from program  participation in the form of higher levels of consumption as well as asset accumulation and better nutrition  for  selfâ€?help  group  participants,  but  not  for  those  who  were  merely  living  in  program  areas  (nonâ€?selfâ€?help  group participants).  Per capita consumption increased for participants by 11 percentage points as compared  to the counterfactual. The increase in consumption and nutritional intake was approximately 15 percent and  the increase in asset endowments was as high as 26 percent. The poorest of the poor also saw increases in  16  nutritional intake and asset accumulation, but only marginally significant impacts on consumption (Deininger  et al. 2009).    Nepal  Poverty  Alleviation  Fund  (PAF).  Begun  in  2004,  Nepal’s  PAF  was  a  response  to  the  country’s  high  poverty levels and deep inequalities, especially among ethnic minorities and lowerâ€?caste communities.  The  objective of PAF was to improve living conditions, livelihoods, and empowerment among the rural poor, with  particular  attention  to  groups  that  were  traditionally  excluded  due  to  their  gender,  ethnicity,  caste,  or  location.  The program supports smallâ€?scale village infrastructure and services (microâ€?irrigation, link roads,  culverts/bridges,  microâ€?hydro,  water  supply,  sanitation,  schools  and  health  posts),  and  income  generation  activities (agriculture, livestock, cottage industries, trade and services). According to the final project report  for  the  first  phase,  72  percent  of  the  project  resources  went  into  income  generation  and  17  percent  to  community infrastructure.    The evaluation was conducted between late 2007 and early 2010 using a randomized phaseâ€?in approach in  which  certain  localities  were  randomly  assigned  for  earlier  intervention  than  others.    The  evaluation  used  two  rounds  of  survey  data  and  differenceâ€?inâ€?difference  combined  with  an  instrumental  variable  estimation  method.  The estimated net program impact on real per capita consumption growth was 19 percent for PAF  participants.  Other  impacts  included  a  19  percentage  point  decline  in  the  incidence  of  food  insecurity  (defined as selfâ€?reported food sufficiency for six months or less)  (Parajuli et al. 2012).     SOUTHEAST ASIA  Indonesia Kecamatan Development Program Phase II (KDP2).  KDP began at the time of the 1998 Asian  economic financial crisis and downfall of the Suharto authoritarian regime. The objectives of KDP2 were to  (a)  support  participatory  planning  and  development  management  in  villages;  (b)  support  a  broad  construction program of social and economic infrastructure in poor villages; and (c) strengthen local formal  and informal institutions by making them more inclusive, accountable, and effective at meeting villagers’ selfâ€? identified development needs.  Most of KDP2’s subproject investments supported rural infrastructure (roads,  bridges, irrigation, and water and sanitation), with a smaller percentage (approximately 25 percent) toward  microâ€?credit activities.    In Indonesia, using national household survey statistics data, a study of KDP Phase I (1998–2002) found that  KDP  had  a  significant  impact  on  per  capita  consumption  in  comparison  with  control  groups,  and  that  the  longer  communities  participated  in  the  program,  the  more  benefits  increased  (Alatas  2005).    The  impact  evaluation for KDP Phase II covered the period from 2002 to 2007, using a differenceâ€?inâ€?difference approach  with propensity score matching.  The impact evaluation found that real per capita consumption gains were 11  percentage  points  higher  among  poor  households  in  treatment  areas  compared  with  control  areas.  The  program  was  most  effective  at  reaching  poor  households  and  households  in  poorer  subdistricts.    The  proportion of households moving out of poverty in poor subdistricts was 9.2 percent higher in KDP2 areas  compared with control areas, and vulnerable households near the poverty line were less at risk of falling into  poverty as a result of participation in the program.  However, households in less poor subdistricts saw either  no benefit or negative impacts. These findings reinforced the view that KDP2 is more beneficial in poor and  5 remote areas (Voss 2008).                                                                      5   A followâ€?up survey covering 2007–09 for KDP/PNPMâ€?Rural is currently under peer review and will be completed in midâ€?2012.   17  Indonesia Kecamatan Development Program/Badan Reintregrasi Aceh (KDP/BRA). The aforementioned  program, KDP, has been working in the northwestern province of Aceh since 1998, even through the years of  conflict  and  martial  law  in  that  province.  Immediately  after  the  tsunami  in  2005  and  peace  agreement  in  2006, the government adapted the KDP program to meet the needs of Aceh.  A new program was developed  called  KDP/BRA  to  support  postâ€?conflict  reintegration  and  recovery.    Unlike  KDP  implementation  in  other  parts  of  the  country,  89  percent  of  the  funds  went  toward  economic  activities  for  individuals  or  groups,  whereas 10 percent were for public infrastructure investments.  The program only lasted one year, between  August  2006  and  2007  (the  planned  second  phase  of  the  program  was  not  implemented).    The  impact  evaluation, using propensity score matching, found that village heads reported an 11 percentage point decline  in the share classified as poor by village heads, as compared with controls (Barron et al. 2009). Funds led to  increased ownership of assets among households in general and conflict victims in particular.  Land use for  conflict victims doubled.  There were no increases in employment levels.     Indonesia  Urban  Poverty  Program  II  (UPP2).    This  urban  CDD  program  began  in  1999  and  is  still  going.  Similar  to  its  rural  sister,  KDP,  UPP  began  in  the  aftermath  of  the  East  Asian  financial  crisis  and  political  transition  in  the  country.    The  objectives  of  UPP2  were  to  (a)  establish  or  support  representative  and  accountable community organizations that could provide services to the urban poor and increase the voice of  the  poor  in  public  decision  making;  (b)  make  local  government  more  responsive  to  the  needs  of  the  poor  through increased cooperation with community organizations; and (c) improve services for the urban poor,  including financial services, social services, and infrastructure.    The  impact  evaluation,  using  differenceâ€?inâ€?difference  matching  methodology,  found  that  UPP2  had  no  statistically significant impact on the welfare of the population living in the project areas as measured by per  capita consumption, access to credit, or assets.  This evaluation raises questions about how to adapt the CDD  model to particular urban settings.     Philippines KALAHI­CIDSS (KC). Started in 2002 to tackle the country’s persistently high rates of poverty and  the  lack  of  community  empowerment  in  the  development  process,  the  Philippine’s  KC  program  aimed  to  strengthen the participation of local communities in village governance, and develop their capacity to design,  implement,  and  manage  development  activities  that  reduce  poverty.  Subproject  funds  have  been  used  primarily to support rural infrastructure such as roads, small bridges, water systems, school buildings, health  stations,  and  daycare  centers.  The  program  undertook  evaluation  survey  rounds  in  2003,  2006,  and  2010,  making this quasiâ€?experimental evaluation the longest in this study in terms of measuring impacts after seven  years  of  implementation.  The  impact  evaluation  found  that  the  project  had  a  positive  impact  on  household  consumption,  with  per  capita  consumption  increasing  by  about  5  percent  as  a  result  of  the  project.  Those  impacts  were  stronger  for  households  that  were  classified  as  poor  in  2003  and  for  households  living  in  villages or barangays that received one or more subprojects.  Labor force participation also increased with a  four  percentage  point  increase  in  labor  force  participation  compared  to  control  groups.    The  evaluators  hypothesize  that  these  welfare  changes  came  about  through  diversification  of  sources  of  income  and  improved production practices.   Farmers in project areas were more likely to sell their produce—an increase  of 14 percentage points over control areas (APPC 2011; Labonne 2011).      18  AFRICA  Senegal National Rural Infrastructure Program (PNIR). The specific objectives of the PNIR when it began  in 2000 were to (a) improve decentralized local governance and local capacity; (b) establish participatory and  decentralized mechanisms for selecting, funding, and implementing rural community investment programs;  (c) strengthen the national institutions supporting decentralization; and (d) implement basic infrastructure  in  a  selected  number  of  communities.    The  government  sought  to  improve  living  conditions  in  rural  areas,  revitalize  the  rural  economy,  and  strengthen  decentralization  as  a  means  to  promote  socioeconomic  development.    Block  grants  under  PNIR  were  used  for  water  supply  projects,  schools,  health  centers  and  health  posts,  livestock  projects,  promotion  of  women’s  activities,  agriculture,  youth  and  sports,  and  commerce.      Increases  in  household  expenditure  per  capita  were  reported for  those  residing  in  a  village  that  received a  completed agricultural or education infrastructure subproject, but not for subprojects in the areas of health  and  potable  water.    A  completed  project  in  the  village  increased  per  capita  household  expenditures  by  65  percent,  with  this  effect  being  highly  significant.  The  large  effect  could  be  due  partially  to  the  fact  that  the  households are particularly poor, equivalent of $0.23 per household member per day.  These results appear  to be driven by agricultural (incomeâ€?generating) projects and to some degree increased primary educational  opportunities (Arcand and Bassole 2007).    Sierra  Leone  GoBiFo  (“Forge  Aheadâ€?)  Project.  Begun  in  2005  following  the  tumultuous  period  of  armed  conflict  in  the  country,  GoBiFo  was  a  fourâ€?year,  $2.5  million  project  funded  through  the  Japanese  Social  Development Fund and Italian government grants.  GoBiFo is the smallest project, in budget terms, included  in  this  study.  The  project’s  development  objectives  were  to  strengthen  social  capital  by  enhancing  the  capacity  of  villages  and  local  governments  to  design  and  implement  strategic  development  plans  at  village  and ward levels; and to enhance accountability of decision makers through mechanisms that fostered open  and  transparent  governance.    The  project  funded  a  range  of  activities  for  skills  training  and  incomeâ€? generation activities, as well as local public goods infrastructure.      The  randomized  impact  evaluation—covering  a  fourâ€?year  period  from  2005  to  2009—revealed  strong  positive  impacts  on  household  living  standards  and  community  market  activity  in  treatment  areas  as  compared with control areas.  Treatment communities had a larger stock of local public goods that were of  higher  quality  than  in  control  areas,  and  beneficiaries  were  better  off  in  terms  of  household  assets  and  increased market activity—such as more petty traders and goods on sale—in their villages. Given the extreme  poverty,  recent  recovery  from  civil  war,  and  prevalence  of  corruption  in  the  country,  the  project’s  achievements were considered relatively good.  Leakage of project resources appears minimal (Casey et al.  2011).    Summary for Poverty Reduction Impacts. Nine projects reviewed in this study reported on income poverty  impacts.    Out  of  the  nine,  seven  had  statistically  significant  positive  impacts  on  household  living  standards  and welfare.      3.  Poverty Targeting  We now turn to the question of whether or not CDD programs reach poor areas and poor households.  And do  they  reach  other  targeted  vulnerable  groups,  such  as  ethnic  minorities,  conflict  victims,  or  those  with  low  education or health levels?  19    In answering the targeting questions, it is important to remember that most CDD programs invest in public  goods  such  as  infrastructure  and  services—roads,  bridges,  schools,  and  health  centers—that  benefit  the  broader  community;  that  is,  they  benefit  both  the  poor  and  nonâ€?poor.  In  the  case  of  CDD  economic  infrastructure—such  as  roads,  bridges,  and  markets,  for  example—the  community  may  choose  to  promote  economic  growth  for  the  entire  area  so  that  everyone  benefits,  as  opposed  to  more  narrowly  targeted  assistance programs such as individual household safety nets.  That being said, we need to examine whether  these programs have benefited more poor than nonâ€?poor households and individuals.      Positive  findings  were  evident  in  Nepal’s  PAF,  India’s  Andhra  Pradesh,  and  the  Philippine’s  KALAHIâ€?CIDSS.   Other  projects  had  positive  to  mixed  findings,  including  Indonesia’s  three  programs,  KDP2,  KDPâ€?BRA,  and  PNPMâ€?Generasi,  Afghanistan’s  NSP2  (for  women),  Senegal  PNIR,  and  the  social  funds  in  Bolivia,  Honduras,  Nicaragua, Peru, and Zambia.  In general, the geographic distribution of program funds from the central level  was  proâ€?poor,  especially  in  those  programs  that  used  poverty  maps  and  the  latest  survey  information  to  target poor areas of the country. In terms of secondâ€?level household targeting, much of the data indicates that  benefits  were  concentrated  among  the  poor,  with  poorer  households  more  likely  to  benefit  from  CDD  investments than betterâ€?off households.  The poor accounted for a greater share of beneficiaries than the nonâ€? poor; the poorest of the poor were wellâ€?represented among project beneficiaries.  However, in most cases of  the social funds, the distribution of beneficiaries at the household level was consistent with national poverty  levels,  with  little  positive  discrimination  of  greater  representation  of  the  poor  except  for  Peru.  Targeting  results also varied by type of investment.     The evidence on targeting is as follows:    Afghanistan  National  Solidarity  Program.  As  mentioned  above,  at  the  time  of  the  interim  evaluation  in  2009,  many  subprojects  were  not  yet  fully  completed,  so  the  interim  evaluation  focused  primarily  on  the  socialization  and  preparation  phases  of  the  program  and  how  the  creation  of  community  development  councils and the selection of development projects affected political and social attitudes and outcomes. Since  its beginning, a special focus of NSP has been the involvement of women in the program.  The 2009 midline  survey report found that the program increased the engagement of women across a number of dimensions of  community life, while also increasing respect for senior women in the village and making men more open to  female participation in local governance. NSP also increased the availability of support groups for women and  reduced  extreme  unhappiness  among  women.  With  respect  to  access  to  medical  care  and  schooling,  the  program’s impacts, at this stage, appear limited to female villagers. Women’s access to professional medical  services  appears  modestly  improved  by  NSP,  but  there  is  no  evidence  of  an  improvement  for  villagers  generally.  Some  evidence  exists  that  NSP  increases  girls’  school  attendance  rates.  NSP  results  in  increased  involvement  by  women  in  incomeâ€?generating  activities,  but  there  is  no  evidence  of  impacts  on  asset  ownership by women,  or on the involvement of women in household decisions.  The next round of impact  evaluation,  currently  under  way,  will  provide  more  complete  information  once  subproject  activities  are  completed.    India  Andhra  Pradesh.  The  poorest  of  the  poor  experienced  nutritional  gains  and  increased  their  level  of  asset accumulation but not consumption. There was approximately a 15 percent increase in nutrition intake,  and  the  percentage  change  in  assets  was  approximately  50  percent  as  compared  with  the  counterfactual  (Deininger et al. 2009).  According to the project’s completion report, one reason the project was successful in  20  targeting was the use of a statewide revised list of the poorest of the poor and the poor households, using an  objectively triangulated participatory and transparent approach (World Bank 2007c).    Nepal  Poverty  Alleviation  Fun.  The  project  objective  particularly  targeted  groups  that  have  traditionally  been excluded due to gender, ethnicity, caste, or location. The project provided grants for incomeâ€?generation  activities,  selected  on  a  demandâ€?driven  basis,  to  groups  of  poor  and  excluded  people  based  on  objective  criteria  including  ethnicity,  caste,  gender,  and  poverty  level.  In  PAF,  the  targeting  of  disadvantaged  groups  seems to have been particularly effective.   The impact evaluation showed a higher decline in food insecurity,  24  percentage  points  for  disadvantaged  households  such  as  those  from  the  dalit,  jananati,  and  other  caste/ethnic  groups  (Parajuli  et  al.  2011).      The  successful  targeting  is  attributed  to  the  way  the  project  manages  the  identification  of  community  organization  members.    Partner  organizations  work  with  communities  using  participatory  methods  to  identify  the  poor  and  socially  disadvantaged.  The  selection  of  community organization members was based on a participatory wellâ€?being ranking that used household food  sufficiency as a proxy for poverty category (World Bank 2009a).    Indonesia Kecamatan Development Program Phase II.  The IE found that real per capita consumption gains  were 11 percentage points higher among poor households in treatment areas compared with control areas.  The  program  was  most  effective  at  reaching  poor  households  and  households  in  poorer  subdistricts.    The  proportion of households moving out of poverty in poor subdistricts was 9.2 percent higher in KDP2 areas  compared with control areas, and vulnerable households near the poverty line were less at risk of falling into  poverty as a result of participation in the program.  However, households in less poor subdistricts saw either  no  benefit  or  negative  impacts.  This  reinforces  the  view  that  KDP2  is  more  beneficial  in  poor  and  remote  areas.    Femaleâ€?headed  households  and  households  where  the  head  of  the  household  lacked  primary  education  saw  no  impacts  in  real  per  capita  consumption.    However,  contrary  to  the  consumption  results,  household heads with relatively less education experienced large gains in access to outpatient services due to  KDP2. Given the small amount of project funds used for construction of health facilities, the biggest factors  contributing  to  improving  health  access  may  have  come  from  road  construction  and  rehabilitation  (Voss  2008).    Indonesia KDP/BRA. This program was designed to assist conflict victims across Aceh province as part of the  reintegration  program  emerging  from  the  Helsinki  Peace  Agreement,  which  brought  to  an  end  a  30â€?year  conflict between the separatist Free Aceh Movement (GAM) and the government of Indonesia. This program  provided additional funds channeled through the existing Kecamatan Development Program.      The  program  was  relatively  successful  at  reaching  large  numbers  of  direct  beneficiaries,  including  both  conflict  victims  and  others.    An  estimated  530,000  individuals  lived  in  households  that  directly  received  assistance.    There  was  widespread  participation  in  the  program  and  meetings,  with  poorer  and  femaleâ€? headed households as likely to attend as others. Across the province, a larger proportion of conflict victims  than  nonâ€?victims  received  support.  Withinâ€?village  targeting  was  relatively  blunt.    in  that  conflict  victims  tended  not  to  do  much  better  than  nonâ€?victims  (Barron  et  al.  2009).    However,  in  volatile  and  widespread  poverty situations such as Aceh, many in the development community would argue that benefits should have  been  widespread  and  targeting  the  program  too  narrowly  would  be  exclusionary  and  potentially  explosive  (see section below on conflict).    Indonesia  PNPM  Generasi  Sehat  dan  Cerdas  Program—PNPM­Generasi,  or  the  National  Program  for  Community  Empowerment—a  Healthy  and  Smart  Generation.  This  program  began  in  2007  as  a  pilot  21  under the Kecamatan Development Program and can be viewed as a hybrid of the traditional CDD model and  a conditional cash transfer model to households. The pilot began as a means of testing a community approach  to  reaching  lagging  education  and  maternal  and  child  healthcare  outcomes  that  were  not  being  fully  addressed through the KDP mechanism. Communities were provided with facilitation and block grant funds  to  reach  12  education  and  health  targets  related  to  improving  primary  and  junior  secondary  school  enrollment  and  attendance  as  well  as  child  and  maternal  healthcare.    The  objective  of  the  program  was  to  improve  certain  education  and  health  outcomes  in  these  areas  As  an  added  innovation,  community  grant  amounts were incentivized based on village performance in attaining the 12 target indicators.    The  randomized  impact  evaluation  measured  effects  in  2008–09  (midâ€?term)  and  2009–10,  with  the  final  round of evaluation taking place after 2.5 years of implementation.  The evaluation found that PNPMâ€?Generasi  had  the  greatest  impact  in  areas  with  low  baseline  health  and  education  indicators.  There  was  robust  evidence that the program was more effective in areas where baseline service levels were lower. On average,  the program was about twice as effective in areas at the 10th percentile of service provision (very low health  and education coverage) at baseline as it was on average. The greater impacts in areas with a lower baseline  appear more prominently in the final evaluation survey than the interim results, with stronger improvements  found  in  education  indicators  in  these  areas.  However,  these  improvements  in  health  and  education  indicators in areas with low baseline coverage did not appear to have resulted in improving longâ€?term health  and  education  outcomes—improved  test  scores,  for  example—in  these  areas.  Furthermore,  the  greater  impacts  observed  in  health  and  education  indicators  were  not  simply  correlated  with  preâ€?project  levels  of  poverty, but instead were driven by the level of health and education indicators in the area.      In terms of household targeting, in general there was little difference in impact among the relatively betterâ€? off  (top  three  quintiles)  and  the  poor  (bottom  two  quintiles)  households  in  both  health  and  education  indicators  in  the  interim  evaluation  and  health  indicators  in  the  final  evaluation.  However,  the  impact  on  education in the final evaluation was largely seen among the poor households, though the difference in total  average  standardized  effects  is  not  statistically  significant.  Specifically,  among  the  poorest  two  quintiles,  primary school participation and attendance among 7â€?toâ€?12 yearâ€?olds increased by 2 percentage points, a 1.8  percent  increase  in  participation  and  2.3  percent  increase  in  attendance,  compared  to  control  areas.  Ageâ€? appropriate  junior  secondary  school  participation  among  13â€?toâ€?15â€?yearâ€?olds  improved  by  7.5  percentage  points,  an  11  percent  increase  compared  to  control  areas,  whereas  there  were  no  impacts  among  the  relatively betterâ€?off households (Olken et al. 2011).    Indonesia  Urban  Poverty  Program  II.    As  mentioned  above,  UPP2  had  no  impact  on  welfare;  however,  access  to  sanitation  for  households  from  the  poorest  quintile  saw  a  10  percentage  point  increase.    As  for  membership  in  community  and  credit  groups,  the  evaluation  shows  that  the  more  educated,  affluent,  and  officialâ€?connected are more likely to get elected into the key project community organizations responsible for  allocating UPP2 resources.  The project credit groups show a similar pattern, although the members are more  likely to be female. They are generally more educated, richer, and more likely to be employed than women in  the general population (Pradhan et al. 2009).    Philippines KALAHI­CIDSS.   Labonne and Chase (2009) found that the poorest villages were more likely to  be  prioritized  during  the  multiâ€?village  decisionâ€?making  forums.  Surprisingly,  the  authors  also  found  that,  controlling for poverty, more unequal villages were more likely to receive funding. They hypothesized that in  more unequal villages, the elected village leader was more likely to override community preferences, and to  influence interâ€?village competition such that project resources flowed to their villages.  As part of the 2010  22  evaluation round, the impact evaluation found that positive impacts on per capita consumption were stronger  for households that were classified as poor in 2003, but had a small negative impact on nonâ€?poor households.   The greatest impacts were found where poverty among households and communities was the highest. This  means that benefits in this case were not captured by local elites (APPC 2011; Labonne 2011).      Senegal PNIR.  In terms of targeting and who receives benefits, the Senegal case points to some evidence of  political  capture.    The  evaluation  found  that  the  probability  of  a  village  receiving  a  subproject  increased  significantly in those villages with a resident who was a member of the majority party on the Conseil rural— an  interâ€?village  council  that  decides  on  subproject  funding.  The  study  found  that  village  chiefs  and  subregional politics play a significant role in determining which villages receive projects and which do not.     Interestingly,  the  number  of  female  villagers  on  the  Conseil  rural  significantly  increased  the  likelihood  of  a  village receiving a completed project.  This result suggests that while the elite capture phenomenon may be  present, it was tempered either by the “voiceâ€? given to women in the CDD decisionâ€?making process, or there  could  have  been  coâ€?optation.  There  is  no  quantitative  or  qualitative  information  to  further  explore  the  dynamics behind this finding.    Despite  the  political  biases  mentioned  above,  the  evaluation  found  that  the  poor  were  still  the  biggest  beneficiaries of the program and that residing in a PNIRâ€?eligible area significantly reduced the prevalence of  underweight  and  stunted  children,  with  this  effect  being  particularly  pronounced  for  children  residing  in  poor households (Arcand and Bassole 2007).    Tanzania Social Action Fund II (TASAF2). Baird et al. (2009) examined the targeting issue for the Tanzania  Social  Action  Fund  II.    TASAF  subprojects  target  three  main  beneficiary  groups:  (1)  poor  communities  (improvement  of  social  services  and  infrastructure),  (2)  foodâ€?insecure  households  (public  works  programs  where  beneficiaries  receive  cash  for  work),  and  (3)  vulnerable  groups,  such  as  the  elderly,  people  with  disabilities,  widows,  orphans,  and  those  affected  by  HIV/AIDS.  As  with  other  social  fund  programs,  the  authors found that the centralized funding allocation to districts was proâ€?poor.  However, the withinâ€?district  targeting  at  best  was  neutral,  making  overall  targeting  performance  for  the  program  mildly  proâ€?poor.    The  authors  also  found  that  wards  with  higher  civic  participation  (in  terms  of  voter  registration  and  turnout)  benefited much more from TASAF.  Furthermore, the internal process of application for funds within districts  generated an initial pool of projects that was strongly regressive. While beneficiary households were slightly  poorer  than  the  average  eligible  household,  they  were  also  much  more  likely  to  be  civically  and  politically  active, related to the village leaders, and be beneficiaries of other welfare programs.  These results point to  the  importance  of  political  involvement  and  widespread  access  to  program  information  in  order  for  communities  to  mobilize  and  file  proposals.    Those  who  are  marginalized  or  poorly  educated  are  likely  to  have less program awareness and have a more difficult time navigating the application process.    Zambia Social Recovery Project II. The Zambia Social Fund relied on selfâ€?targeting and outreach activities to  reach the poor.  The social fund reached poor households, particularly in rural areas. However, urban social  fund  beneficiaries  were  relatively  better  off  than  average  urban  households,  revealing  that  urban  selfâ€? targeting was less effective. The evaluators believe that this result in the urban areas was due to the social  fund  being  the  executing  agency  for  the  Ministry  of  Education’s  urban  school  rehabilitation  program.  The  urban  implementation  arrangements  thus  did  not  follow  the  normal  social  fund  selfâ€?targeting  procedures  (Chase and Sherburneâ€?Benz 2001).      23  In another study, De Janvry et al. (2009) looked at whether ZAMSIF project allocations across either districts  or  wards  became  more  progressive  or  regressive  as  the  project  decentralized  over  its  three  phases  of  implementation, covering a time span of approximately 16 years. The study found that the decentralization of  the  program’s  functions  to  more  administratively  capable  local  districts  led  to  more  progressive  targeting  across wards, mildly so at the national level and distinctly so within districts. The authors also looked at the  interplay of politics with resource allocations as the program decentralized.  They found that local votes were  increasingly  rewarded  by  the  allocation  of  local  projects,  and  local  projects  were  increasingly  rewarded  by  electoral  support  for  incumbent  politicians.    Thus  the  decentralization  trend  made  local  politics  more  important  in  influencing  resource  allocations  and  in  rewarding  elected  officials  for  delivering  local  public  goods.  However, as with the Senegal case, the political influence did not mean that resources were allocated  to richer areas; the data shows wards that most voted for the majority party of the district tended also to be the poorest wards. Thus, the allocation of projects still favored relatively poor wards.     Armenia Social Investment Fund. The impact evaluation for the Armenia Social Investment Fund relied on a  nationally  representative  household  survey,  oversampled  in  areas  where  the  social  fund  was  active.    Using  propensity and pipeline matching techniques, it evaluated the household effects of rehabilitating schools and  water systems.  ASIF households were on average less well off than other households. However, the targeting  of  ASIF  resources  was  relatively  neutral  with  regard  to  poverty—slightly  progressive  in  urban  areas,  and  slightly  regressive  in  rural  areas.  The  author  hypothesized  that  one  explanation  for  the  progressive  urban  targeting was that the fund’s focus was on Yerevan, whose population suffered more acutely from economic  dislocation.    The  regressive  rural  targeting  may  result  from  the  difficulties  rural  communities  faced  in  providing the required 10 percent community contribution (Chase 2002).    In the crossâ€?country analysis of social funds in Armenia, Bolivia, Honduras, Nicaragua, Peru, and Zambia,  Rawlings et al. (2003) found that the geographic distribution of social fund spending was progressive in all  countries,  with  poor  districts  receiving  more  per  capita  than  wealthier  districts.  In  fact,  the  very  poorest  districts  received  shares  exceeding  their  shares  of  the  population.  Overall,  geographic  distribution  favored  the poor more in the countries that use poverty maps to target resources to poor areas. In terms of household  targeting, in all six countries, the data show that social fund benefits were concentrated among the poor, with  poorer households more likely to benefit from a social fund investment than betterâ€?off households.  The poor  accounted  for  a  greater  share  of  beneficiaries  than  the  nonâ€?poor;  the  poorest  of  the  poor  were  wellâ€? represented  among  project  beneficiaries.    However,  in  most  cases,  the  distribution  of  beneficiaries  at  the  household  level  was  consistent  with  national  poverty  levels,  with  little  positive  discrimination  of  greater  representation of the poor except for Peru (Paxson and Schady 2002). Targeting results also varied by type of  investment, with latrines and health clinics doing better at reaching the poor; investments in education were  proâ€?poor as well. Sewerage projects fared the worst. Comparisons with other programs show that the funds’  geographic  and  household  targeting  compared  favorably  with  that  of  other  targeted  social  programs  and  general social spending in the countries concerned.  Benefits reach the nonâ€?poor largely because of the nature  of  social  fund  investments  which  are  concentrated  on  community  infrastructure  and  services,  so  everyone  has access, poor and nonâ€?poor alike.      Summary  for  Poverty  Targeting  Effectiveness.  The  evidence  across  the  14  programs  that  looked  at  targeting  found  generally  positive  evidence  on  geographical  targeting,  especially  when  poverty  maps  were  used to guide investments to poorer areas. Projects also need to be careful of the interplay between politics  and  the  allocation  of  resources  at  the  local  level,  although  in  the  cases  of  Senegal  and  Zambia,  the  political  links did not seem to affect proâ€?poor allocation of resources.  At the level of household targeting, the evidence  24  is  positive  to  mixed,  and  there  is  some  variation  within  some  programs  (social  funds  and  PNPMâ€?Generasi)  and depending on the types of subproject investments.     4.  Access to and use of services  Evidence  across  the  CDD  programs  is  generally  strong  regarding  positive  impacts  on  access  to  and  use  of  services, especially in health, education, and drinking water.  These impacts for the most part derive directly  from the types of subprojects funded through the project grants.  There are some reported positive impacts in  a few projects on longer term outcomes such as child nutrition improvements and underâ€?ageâ€?five mortality.   Two  programs  tested  for  education  learning  achievement  (Bolivia  Social  Fund  and  Indonesia’s  PNPM  Generasi); both found no impacts on longerâ€?term educational achievement.     Afghanistan’s  National  Solidarity  Program  II  (NSP2).  The  impact  of  NSP  on  services,  infrastructure,  and  utilities  was  mixed  at  the  time  of  the  interim  evaluation,  when  many  subprojects  were  not  yet  completed.   Using  data  from  villages  with  completed  projects,  the  evaluation  identified  a  strong  positive  impact  of  drinking  water  projects  that  used  protected  outlets  and  emphasized  the  availability  of  safe  drinking  water,  with a 20 percentage point increase in usage of protected sources.   Completed electricity projects—although  few  in  number—also  suggested  strong  impacts  on  connectivity  and  usage.    There  were  few  impacts  of  infrastructure projects on the mobility of villagers or irrigation outcomes.  With respect to access to medical  care and schooling, the program’s impacts appear limited to female villagers.      Nepal Poverty Alleviation Fund. For education, school participation among children aged 6â€?toâ€?15 increased  by 14 percentage points for all children in the given ageâ€?group, but girls appear to have benefited even more  with  a  21  percentage  point  net  increase.    The  effects  are  as  strong  among  children  from  disadvantaged  caste/ethnic groups.   The authors hypothesized that the program’s gains in household income/consumption  could have reduced potential constraints of sending children to school.  Community mobilization and social  networking may also have had spillover effects in motivating households toward child schooling.  There were  no significant differences on the health side or for child malnutrition (Parajuli et al. 2012).    Indonesia Kecamatan Development Program Phase II. The proportion of household heads gaining access  to outpatient heathcare was 11.5 percentage points higher in KDP2 areas compared with the control group;  however, there were no impacts on school enrollment rates. At the primary school level, this was most likely  due to the already high levels of enrollment. At the secondary level, the lack of individual panel data reduced  the ability of the study to detect impacts (Voss 2008).    Indonesia PNPM Generasi Program.  Communities were provided with facilitation and block grant funds to  reach  12  education  and  health  targets.  The  final  evaluation  in  2009,  conducted  after  2.5  years  of  program  implementation, found that Generasi had a statistically significant positive impact on the 12 indicators it was  designed  to  address.  Among  the  health  indicators,  the  strongest  improvements  were  in  the  frequency  of  weight checks for young children. The program also increased the number of iron sachets pregnant mothers  received  through  antenatal  care  visits.  These  improvements  were  supported  by  dramatic  increases  in  mothers  and  children  participating  in  village  health  post  (posyandu)  activities  to  receive  the  targeted  maternal, neonatal, and child health services. Childhood malnutrition was reduced by 2.2 percentage points,  about  a  10  percent  reduction  from  the  control  level.  This  reduction  in  malnutrition  was  strongest  in  areas  with  a  higher  malnutrition  rate  prior  to  project  implementation.  Education  indicators  also  saw  some  improvements in the final evaluation, reversing the zero or negative impact found at the interim evaluation.  25  The improvement in education indicators was most notable in the increased school participation rate among  the primary schoolâ€?age group. However, no impacts were found on educational test scores for language and  mathematics,  although  that  was  most  likely  due  to  the  short  timeâ€?span  of  measurement  (2.5  years).    The  program worked best in areas of low education and health indicators (Olken et al. 2011).    Indonesia Urban Poverty Program II. The impact estimates indicate an improvement in access to adequate  sanitation as a result of UPP2, in particular for the poor. Access to adequate sanitation increased by about 3  percentage points for the general population. For households from the poorest quintile, the improvement was  on the order of 10 percent. The program had no impact on clean water.      Philippines  KALAHI­CIDSS.  On  the  health  side,  the  proportion of  households  visiting  a  health facility  when  sick increased.  The change appears to be driven by an increase in the use of public village health stations. The  impacts of access to water were limited; the followâ€?up qualitative analysis indicates that (1) there may have  been systems capacity issues preventing the project from serving all households in the village, and (2) there  may have been some maintenance issues.  For education, there were mixed results.  Program implementation  led  to  an  increase  in  secondary  school  and  college  enrollment  but  a  small  decline  in  elementary  school  enrollment. Given the relatively small amount of investments in school buildings in the villages sampled for  the  evaluation,  this  correlation  might  not  be  the  result  of  the  project.    The  positive  impact  on  secondary  school enrollment resulted in a reduction in enrollment for boys and an increase for girls.      The  project  also  had  a  positive  impact  on  accessibility.  Specifically,  the  project  resulted  in  a  6  percentage  point increase in the proportion of households whose house was accessible yearâ€?round by roads. The changes  are larger in prioritized villages, with a 10 percentageâ€?point increase for each completed subproject. Lastly, as  a  byproduct  of  KALAHIâ€?CIDSS,  the  evaluation  found  that  there  were  now  more  financing  institutions  operating in treatment communities, a 9 percentage point difference between treatment and control, and a 50  percent increase over baseline levels (APPC 2011; Labonne 2011).    Senegal  PNIR.      The  evaluation  found  that  completed  subprojects  significantly  increased  access  to  clean  water  and  health  facilities  and  improved  the  nutritional  status  of  children.  This  effect  is  particularly  important for children living in poor households. Reductions in the prevalence of underweight and stunted  children  were  reported.  Moreover,  the  project  had  effects  in  areas  that  did  not  have  completed  projects;  simply  residing  in  a  PNIRâ€?eligible  area  brought  statistically  significant  benefits  in  terms  of  child  health,  perhaps because of spillovers from neighboring villages that receive completed projects (Arcand and Bassole  2007).    Sierra Leone GoBiFo. Access to local public goods improved; treatment communities had more local public  goods  that  were  of  better  quality  than  in  control  areas.    For  example,  treatment  communities  saw  large  increases in the proportion of villages with functional traditional birth attendant houses, community centers,  latrines,  and  seed  banks.  Also  the  project  had  a  strong  positive  impact  on  the  quality  of  the  materials  and  construction for primary schools, grain drying floors, water wells, and latrines (Casey et al. 2011).    Zambia  Social  Recovery  Project  II.  Compared  with  control  groups,  education  projects  increased  school  attendance and households’ education expenditures to a significant degree. This is not surprising, given that  80 percent of the subprojects were in the education sector.  Health projects primarily involved rehabilitating  health posts; they saw increased use of primary health facilities. Households were more likely to attend these  health providers rather than going directly to hospitals. There were also improvements in social fund areas  26  for  some  types  of  child  vaccinations,  specifically  diptheria  vaccinations.    There  is  limited  evidence  of  increased BCG and polio vaccinations (Chase and Sherburneâ€?Benz 2001).    Bolivia Social Investment Fund II (SIF). Education projects either repaired existing schools or constructed  new  ones,  along  with  providing  additional  new  facilities  such  as  furniture,  blackboards,  and  playgrounds.   Schools  that  received  a  Social  Fund  intervention  benefited  from  significant  improvements  in  infrastructure  (the  condition  of  classrooms  and  an  increase  in  classroom  space  per  student)  and  in  the  availability  of  bathrooms  compared  with  schools  that  did  not  receive  an  SIF  intervention.    There  was  also  an  increase  in  textbooks per student and a reduction in the studentâ€?teacher ratio. But the improvements had little effect on  school enrollment, attendance, repetition rates, or academic achievement. Among the studentâ€?level outcomes,  only the dropout rate reflects any significant impact from the education investments.  Health impacts were  more  pronounced.  The  evaluation found  that the  share  of  women  receiving  prenatal  care   and  the  share of  attended births increased significantly. The project also financed smallâ€?scale potable water systems, and later  supported  training  to  maintain  these  systems.    The  evaluation  showed  a  reduction  in  the  distance  to  the  water source and in one area (Resto Rural) a substantial improvement in sanitation facilities.  Both the health  and  water/sanitation  intermediate  results  appear  to  have  led  to  a  significant  reduction  in  underâ€?ageâ€?five  mortality in the areas served by the health clinics receiving a SIF intervention. (Newman et al. 2002)    Honduras Social Investment Fund III (FHIS). In Honduras, the Social Fund had an effect on school size, but  there  was  no  corresponding  impact  on  enrollment.  The  evaluators  hypothesized  that  this  might  have  been  due to already high enrollment rates (89 percent). The project also reduced ageâ€?forâ€?grade by 15 percent. For  health,  household  survey  data  showed  that  social  funds  investments  led  to  a  10  percentage  point  increase  (from  41  percent  to  51  percent)  in  the  share  of  sick  people  seeking  professional  medical  services;  beneficiaries  sought  treatment  at  the  local  center,  saving  time  and  money.  The  qualitative  work  also  supported the finding of increased utilization, especially for pregnant women and children due to the closer  proximity  of  the  health  center  and  involvement  of  community  members  as  health  volunteers  (ESA  Consultores 2005; World Bank 2000a).    Nicaragua  Emergency  Social  Investment  Fund  II  (FISE).    In  Nicaragua,  the  net  primary  enrollment  rates  was  5  to  10  percentage  points  higher  in  social  funds  than  in  nonâ€?social  fund  communities.  The  impact  on  enrollment was larger for girls. Female enrollment rates were 4 percentage points higher than male rates.  On  the health side, social fund health posts exhibited slightly greater and significant increases in utilization rates.   For  vaccination  coverage,  both  treatment  and  comparison  groups  had  high  overall  coverage  rates  for  diptheria,  pertussis,  tetanus,  and  polio  vaccinations.    Regarding  disease  incidence,  social  fund  communities  had a higher incidence of diarrhea and respiratory infections compared with the propensity score matching  comparison group, but not compared with the other comparison groups. In terms of access to piped water,  social fund investments increased the share of households with access to piped water considerably and there  was  a  significant  decline  in  the  distance  to  the  water  source.    The  evaluation  reported  an  improvement  in  health  outcomes  due  to  the  water  investments,  with  the  incidence  of  stunting  (low  heightâ€?forâ€?age)  falling  from 25 to 14 percent (World Bank 2000d).    Peru  Social  Investment  Fund  (FONCODES).  In  Peru,  the  evaluation  (Paxson  and  Schady  2002)  concluded  that increased school attendance was directly correlated with the social fund resources that an area received,  and the districts that received the most resources for school improvements had achieved the largest gains in  primary  school  enrollment  rates.    However,  in  the  Instituto  Apoyo  study  (2000)  covering  only  rural  areas,  there  was  no significant  difference  in  enrollment  of children  ages  6  to  14.    The  study  did  find  a significant,  27  positive impact on school enrollment among households in extreme poverty, e.g., those living on less than $1  a day; social fund treatment areas raised the probability of being enrolled in school by almost 2 percent.  As  the majority of its education projects were in indigenous communities, the evaluation also explored whether  there were different impacts for indigenous communities.  For indigenous communities, the study found no  impact on enrollment rates, attainment, or absenteeism.    Armenian Social Investment Fund. In the earthquake zones, household spending on primary education and  primary  school  attendance  rose  in  communities  that  had  completed  school  projects,  suggesting  increased  demand for education. In communities with water projects, households reported improvements in access to  water  and  in  water  services.    There  were  no  strong  improvements  in  health  in  those  communities  (Chase  2002).     Summary  of  Results  for  Improving  Use  and  Access  to  Services.    Evidence  across  the  CDD  programs  is  generally  positive  regarding  impacts  on  access  to  and  use  of  services,  especially  in  health,  education,  and  drinking water.  These impacts for the most part derive directly from the types of subprojects funded through  the project grants.  There are some reported positive impacts in a few projects on longer term outcomes such  as  child  nutrition  improvements  and  underâ€?ageâ€?five  mortality.    Only  two  programs,  the  Bolivian  Social  Investment Fund and Indonesia’s PNPM Generasi, measured for education achievement in mathematics and  language  tests,  but  there  were  no  impacts  on  longerâ€?term  learning  achievement.  On  the  limited  nature  of  longerâ€?term outcomes, this may again be a function of the short timeâ€?frames for the evaluations. Thus more  longitudinal studies are needed.    5.  Social Capital and Governance   Social capital is broadly defined by the World Bank as “the norms and networks that enable collective action.â€?   The  theory  is  that,  through  communities  working  together  using  a  CDD  approach,  the  project  builds  trust,  networks, and collective action.  Social interaction and networks, trust and reciprocity, as elements of social  capital,  can  produce  collective  outcomes,  both  beneficial  and  harmful  (Grootaert  and  van  Bastelar  2002).    Impact  evaluations  attempt  to  measure  social  capital  improvements  during  the  intervention  and  whether  there are spillover effects outside of the project sphere.  For example, do communities take collective action  to solve other village problems, not necessarily those related to the CDD project?  To measure social capital,  evaluators generally examine three types of proxy indicators:    ï‚· Memberships in networks and associations. Evaluators look at the density of networks,  involvement  in associations, and group membership.  ï‚· Local  collective  action.  Key  questions  relate  to  how  groups  of  individuals  work  together  to  solve  collective action problems.   ï‚· Trust. Questions look at trust toward other members of the community as well as different levels of  government in terms of decision making and the delivery of services.    Of  the  projects  covered  under  this  study,  only  one  (Sierra  Leone)  had  an  explicit  goal  of  improving  social  capital.  However,  as  a  variable  of  high  interest  to  development  practitioners,  it  is  sometimes  measured  in  impact evaluations.       It is worth noting that the evidence base on social capital and local governance is very limited.  Of the 17 total  projects discussed in this study, only seven looked at various dimensions of social capital formation, and only  28  four asked questions explicitly related to local governance.  The evidence to date indicates that most projects  have limited impact on social capital.    Turning  to  local  governance  aspects,  when  CDD  programs  aim  to  improve  local  governance,  this  usually  refers  to  changes  in  the  way  government  interacts  with  its  citizens  within  the  project  domain  as  well  as  outside  of  project  parameters.    This  refers  to  citizen’s  participation  in  decision  making  and  management,  transparency  of  program  information,  efficient  and  effective  delivery  of  services,  and  accountability  to  citizens.  As proxies for improved local governance, impact evaluations often measure:    People’s attitudes toward various levels of government   ï‚· Participation in public assemblies or meetings inside and outside of the project domain  ï‚· Awareness of project information and other local civic activities   ï‚· Spillover  effects  in  terms  of  the  way  government  officials  and  citizens  approach  and  manage  other  development  programs  and  civic  activities.  For  example,    do  they  use  a  participatory  approach  or  communityâ€?led decisionâ€?making model in other programs?      For governance impacts, there are positive to mixed results.  Only four CDD programs examine explicitly local  governance  issues.    Afghanistan  and  the  Philippines  do  well  across  the  board  on  higher  village  assembly  attendance  (outside  of  the  project);  greater  awareness  of  village  and  civic  issues;  and  improved  attitudes  toward  government.  Sierra  Leone  also  showed  positive  impacts  on  local  leaders’  behavior  in  planning  and  management of activities, which increased community confidence toward local officials. However, Indonesia’s  KDPâ€?BRA  showed  no  improvements  in  local  governance.  The  project  lasted  for  only  one  year,  which  may  explain the lack of impacts.     Afghanistan National Solidarity Program II (NSP2). Based on the midline evaluation, there is weak or no  evidence  that  NSP  affects  levels  of  trust  between  villagers.    NSP  has  no  impact  on  specific  measures  of  community trust or solidarity or on the outbreak of village disputes or tribal feuds.     On  the  governance  front,  there  are  more  positive  results.    By  supporting  community  development  councils  (CDCs), NSP altered the structure of village governance by supporting these newly elected bodies as part of  the  program  and  transferring  some  authority  from  tribal  elders  to  these  councils.  The  program  does  not  appear to affect activities undertaken by customary villager leaders. The program stimulated participation in  local governance by increasing both the frequency and attendance at meetings of the village assembly.  Male  villagers also perceived government actors more favorably, and the program increased connections between  villages and select governmental and nongovernmental institutions.  The interim evaluation also found that  the  project  increased  the  participation  of  women  in  local  governance  and  their  awareness  of  village  leadership  and  local  governance  services.  Women  in  NSP  areas  were  also  found  to  have  increased  the  occurrence  of  meetings  between  female  villagers  and  women  from  other  villages,  as  well  as  district  governments. But there was no impact on the general extent of socialization between female villagers or the  frequency of female villagers leaving their compounds (Beath et al. 2010).     Nepal  Poverty  Alleviation  Fund.    No  significant  differences  were  evident  between  treatment  and  control  groups  on  indicators  associated  with  community/social  capital  (trust,  respect,  relationships  between  different ethnic groups, community disputes, etc.) (Parajuli et al. 2012).    29  Indonesia  KDP­BRA.    In  postâ€?conflict  Aceh,  there  was  little  evidence  that  the  KDPâ€?BRA  program  improved  social  cohesion  or  had  much  impact  on  social  relations  or  trust  in  the  state.  Levels  of  social  acceptance  of  returning groups, reported social tensions, divisions, and conflict were similar between treatment and control  villages.    Overall  involvement  in  associational  life  was  similar  for  program  and  nonâ€?program  areas.    This  could  be  due  to  the  limited  time  frame  of  oneâ€?year  for  the  project,  as  well  as  the  fact  that  the  program  provided mainly private goods, which involved less cooperative work.  Furthermore, there is some evidence  that the program resulted in less acceptance by communities of former combatants.  Conflict victims in areas  that received the program were 18 percent less likely to say they accept exâ€?combatants in all roles in village  life  as  compared  to  control  areas.    While  this  did  not  result  in  increased  social  tensions  or  conflict,  it  does  suggest  that  the  program  was  not  effective  in  building  relations  of  trust  between  victims  and  former  combatants.    Some  reasons  hypothesized  by  the  authors  were  that  exâ€?combatants  who  were  supposed  to  receive payments through another government program tried to get assistance through KDPâ€?BRA as well, and  their attempts to benefit from both programs may have caused resentment among the community (Barron et  al. 2009; Morel et al. 2009).    Related to indicators for local governance, the authors find little evidence that BRAâ€?KDP resulted in changes  in key indicators of attitudes toward government, such as awareness of local government affairs or trust in  government officials.    Indonesia  Urban  Poverty  Program  II.  The  impact  evaluation  showed  no  impact  of  the  program  on  community  participation,  organizational  membership,  or  participation  in  communityâ€?initiated  activities.   However, there was a significant decrease in household contributions to projects and programs (government  or community, excluding UPP) in treatment communities. It could be that the communities’ participation in  UPP  projects  resulted  in  a  decrease  in  participation  in  nonâ€?UPP  projects.  Lastly,  the  evaluation  found  that  while UPP is rather well known in the community and many report to have received benefits, it is not known  for its ability to solve the most important problems the community faces. When discussing the most pressing  problems in the village, and which institutions have dealt with the problems, respondents seldom mentioned  UPP  (Pradhan et al. 2009).    Philippines KALAHI­CIDSS. The Philippines’ evaluation is one of the few studies that show generally positive  impacts  on  social  capital,  as  well  as  spillover  effects.    Whether  or  not  this  is  because  the  evaluation  was  conducted  over  a  longer  time  period  of  seven  years  may  be  one  factor  or  it  could  possibly  be  due  to  the  country  context.    The  final  evaluation—measured  after  seven  years  of  project  interventions—reported  increased  levels  of  trust  within  project  communities  and  increased  memberships  in  organizations.  The  evaluation  showed  a  12.3  percentage  point  increase  in  the  proportion  of  respondents  indicating  that  most  people in their village can be trusted, as well as a smaller but still positive impact on trust in local officials and  in national officials.  Aside from a longer time frame for measurement, one reason why the Philippines project  may  be  having  social  capital  impacts  is  that  the  project  spends  more  time  than  most  on  the  initial  socialization and preparation phase with local government officials and communities, on average 5–6 months  of the cycle; other countries typically spend three months.    For  collective  action  (bayanihan),  the  evaluation  showed  a  decrease  of  2.7  percentage  points  in  treatment  areas.    However,  the  program  also  had  a  positive  impact  on  the  willingness  of  households  to  contribute  money for activities that would benefit the community.  The accompanying qualitative study found that this  decrease  in  collective  action  could  be  due  to  the  increased  opportunity  cost  of  contributing  time  to  such  30  activities  as  a  result  of  increased  returns  of  agricultural  activities.  As  a  result,  households  might  be  substituting away from time contributions and toward monetary contributions.     In terms of improving local governance, the evaluation looked at two variables in relation to participation in  governance:    (1)  attendance  in  village  assemblies,  which  looks  at  the  involvement  of targeted  communities  with local government through increased participation in village assemblies; and (2) understanding of village  affairs. The project appeared to have some positive impacts on villagers’ participation in village assemblies,  with  a  5  percentage  point gain  in  treatment  areas.  The  accompanying  qualitative  data  shows  an  associated  qualitative  change  in  how  the  village  assemblies  were  perceived.    Prior  to  project  implementation,  the  assemblies were, at best, considered avenues for reporting, whereas in project areas, the assemblies tended  to be seen as mechanisms for participation, transparency, and accountability.  The proportion of households  who  are  aware  of  the  income  and  expense  details  of  their  barangay  local  government  unit  also  increased.  However, these increases attenuate over time, with strong impacts during the 2006 midâ€?term evaluation and  smaller impacts by the time of the endline survey in 2010.  This could mean that governance impacts are tied  to project implementation (APPC 2011; Labonne 2011).    Sierra Leone GoBiFo. The evaluation reveals no treatment effects on the standard proxy measurements for  social capital—trust, collective action, groups and networks, inclusion, and information.  There was also no  indication  of  spillover  of  local  norms  or  institutional  practices  outside  the  immediate  project  sphere;  for  example, women were no more likely to speak up in community meetings held after the project ended.        Related  to  local  governance  improvements,  local  government  leaders  were  more  active  in  the  planning,  construction, and oversight of local public goods in treatment communities. The greater interaction appears  to have increased citizen confidence in their local representatives (Casey et al. 2011a).    Zambia  Social  Recovery  Project  II.  The  evaluation  shows  that  the  social  fund  had  important  impacts  on  social  capital,  working  in  rural  areas  to  bring  communities  together.  In  rural  households  where  the  social  fund  operated,  households  were  more  likely  to  have  participated  in  the  project.  Subjective  measures  of  community  togetherness  increased.  Further,  those  communities  were  more  likely  to  undertake  other  community initiatives, such as building new health posts and other new schools.  However, urban areas with  social funds saw no such social capital improvements (Chase and Sherburneâ€?Benz 2001).    Armenia  Social  Investment  Fund.    Communities  that  had  completed  a  social  fund  project  were  less  likely  than comparison groups to complete other local infrastructure projects— for example, road works, a piped  water  system,  or  health  facility  rehabilitation—suggesting  that  social  capital  was  expended  in  these  early  projects.    However,  communities  that  joined  ASIF  later  and  had  not  yet  completed  their  projects  reported  more collective action (Chase 2002).    Summary for Social Capital and Governance Impacts.  Overall, the evidence shows positive to mixed results  for  local  governance  in  terms  of  attitudes  toward  governance,  participation  in  other  village  assemblies,  awareness  of  project  information  and  other  local  civic  activities,  and  spillover  effects  in  terms  of  the  way  government officials and citizens approach and manage other development programs and civic activities.  For  social  capital,  eight  projects  measured  impacts  related  to  greater  trust,  association,  and  collective  action.   There was very little positive impact on social capital.  Results for the Philippines, Zambia, and Armenia were  mixed,  while  other  programs  such  as  Sierra  Leone,  Indonesia’s  KDPâ€?BRA  and  UPP,  Nepal,  and  Afghanistan  showed no impacts.  31  6.   Conflict  CDD  projects  exist  in  numerous  countries  affected  by  violent  conflict.    Many  governments  use  the  CDD  approach  to  deliver  services  more  rapidly  to  its  citizenry  and  to  build  stateâ€?society  links  through  more  inclusive  decisionâ€?making  processes.    However,  given  these  very  difficult  contexts,  positive  results  are  not  always  guaranteed.    In fragile  situations  where  security  is problematic,  goods  may  not  be  readily  available,  technical  expertise  may  be  limited,  and  where  the  state’s  authority  may  be  in  question,  implementing  any  program could be challenging. This study includes five evaluations examining impacts on violent conflict: two  fragile and conflictâ€?affected countries (Afghanistan and Sierra Leone) and two countries (the Philippines and  Indonesia)  dealing  with  more  localized  conflictâ€?affected  situations.    Not  surprisingly,  there  is  no  impact  on  macro  levels  of  violence  except  with  the  MILF  rebel  group  in  Mindanao,  Philippines.  At  the  more  micro,  village level, however, Indonesia shows some improvements over time in group relations.      Afghanistan NSP. Measured after two years of implementation, NSP had no impact on specific measures of  community  trust  or  solidarity  or  on  the  outbreak  of  village  disputes  or  tribal  feuds.  There  are  very  limited  effects on the prevalence of conflict and perceptions of safety and security (Beath et al. 2010).    Indonesia  KDP.  One  evaluation  uses  reports  of  violent  conflicts  in  local  newspapers  in  two  mediumâ€?level  conflict provinces to assess the impact of KDP (Barron et al. 2006). They find little evidence of a project effect  on violent conflict at an aggregate level or a direct positive impact on nonâ€?projectâ€?related violence at the local  level.  However,  they  did  find  that  KDP  significantly  contributed  to  improvements  in  intergroup  relations.  Across  a  range  of  different  identity  cleavages,  the  study  finds  KDP  helped  contribute  to  improvements  in  intergroup relations.  Improvements in the quality of group relations grew larger over time, with villages that  have had KDP for four years generally showing greater improvements than those that have had the project  for  shorter  periods.  The  study  also  showed  that  KDPâ€?related  conflicts  were  far  less  likely  to  escalate  into  violence  than  were  conflicts  related  to  other  development  projects.  The  reasons  for  this  included  the  presence of effective complaints channels to defuse tensions before they escalated.    Indonesia KDP­BRA.   Barron et al. (2009) found that after one year of project implementation, there were no  impacts  on  social  cohesion,  and  in  fact  there  was  some  evidence  that  KDPâ€?BRA  was  associated  with  less  acceptance of exâ€?combatants by conflict victims in project areas, although there was no evidence that these  tensions escalated to violence. The authors submit that some of these findings were not surprising and the  short duration of the intervention most likely limited the gains in social cohesion that may have been possible  if the program had been implemented for multiple cycles.    Philippines  KALAHI­CIDSS.    A  recent  study  by  Arcand  et  al.  (2011)  looking  at  the  interplay  between  the  incidence and levels of conflict in 2003 and 2006 and the ongoing conflicts created by two rebel groups, the  New People’s Army (NPA) and the Moro Islamic Liberation Front (MILF) in the Philippines.  These two groups  maintain  very  different  ideologies.    The  MILF,  started  in  1981  with  the  objective  of  establishing  an  independent  Islamic  state  in  Mindanao,  are  much  more  concerned with  rectifying  grievances  regarding  the  economic situation of Muslims in Mindanao and their exclusion from the political process. The NPA, on the  other hand, began in the early 1970s as the armed wing of the Community Party of the Philippines and aims  to  overthrow  the  government  through  guerilla  warfare  and  seize  control  of  the  country.  It  is  generally  perceived that MILF’s demands can be accommodated through various institutions, whereas the NPA is intent  on overthrowing the entire government.      32  Using local newspaper reports of conflict (assessing a rather substantial radius of 100 kilometers as a basis  for analysis), a nationally representative household survey, and budget data on municipalities, the evaluation  authors  found  that  the  KALAHIâ€?CIDSS  Project  led  to  a  decline  in  MILFâ€?related  events  and  to  an  increase  in  NPAâ€?related  events.    The  authors  hypothesized  that  the  decrease  in  MILFâ€?related  events  could  derive  from  the project having increased the sense of inclusion of the Muslim population in local decision making and a  reduced  sense  of  grievance  toward  the  central  government.  The  NPA,  however,  perceives  the  project  as  a  threat that might undermine their popular support. The increase of violence by the NPA could be interpreted  as an attempt to prevent the government from implementing an aid project that might reduce grievances.  Or  it could be a reflection of the “rentâ€?seekingâ€? nature of the NPA. Much depends on the goals and motives of the  6 rebel group operating in the contested area.    This is one of the few studies in the conflict and development  literature that examines the different impacts on conflict occurrence depending on the ideology of the rebel  groups  involved.    It  highlights  the  importance  of  understanding  the  nuances  of  local  conflict  dynamics  and  7 how a development program can exacerbate or lessen conflict in certain areas.   It is not clear, however, from  the  development  literature  or  the  impact  evaluations  if  the  interventions  are  due  to  CDD  approaches  in  particular or would occur for any type of governmentâ€?led intervention in conflict areas.    Sierra Leone GoBiFo.  The evaluation reveals no impact on the level of crime and conflict in communities or  the mechanisms through which conflicts are resolved (Casey et al. 2011a).    Summary of CDD impacts on violent conflict. In conclusion, development programs in general must operate  cautiously  in  conflict  and  fragile  situations.  CDD  programs  are  no  exception.  They  can  attract  conflict  by  introducing competition for funds, or bringing more development funds into a community.  But in some select  cases, they can also address longâ€?standing grievances of exclusion and of a nonâ€?responsive state, e.g., MILF in  the Philippines, or introduce community mechanisms for mediating burgeoning conflicts (Indonesia).      Summary for Results Section    Overall, the evidence shows generally positive results on poverty reduction and access and use of services;  positive to mixed results on targeting; mixed results for local governance; and very little positive impact on  social capital and conflict resolution.      In terms of counterfactuals, none of the impact evaluations compare the outcomes of CDD investments with  those  of  alternative  interventions  such  as  direct  central  or  local  government  service  delivery.    Some  cost  comparisons are made in Section VII, but not for outcomes.  Part of the reason for this stems from the fact that  these CDD/social fund programs grew as a response to the failure of decades of topâ€?down, centrally driven  service  delivery,  so  governments  do  not  feel  that  they  need  to  experimentally  test  this  hypothesis.  They  already  had  decades  of  experience  as  testaments  to  the  flaws  in  this  model.  The  second  reason  is  partially  cost. The operational implications and sample sizes to test “topâ€?downâ€? versus “bottomâ€?upâ€? CDD approaches  could  be  fairly  large  and  costly.  Nevertheless,  running  a  randomized  experiment  comparing  these  different  approaches along the dimensions discussed here would be valuable.                                                                        6  This discussion was informed by exchanges with one of the evaluation coauthors and the WB Philippines office.    7 Crost and Johnston (2010) also report on conflict incidence in the KALAHIâ€?CIDSS program  but the authors rely primarily on Armed  Forces of the Philippines data between 2001 and 2008, and there has been a long debate in the Philippines regarding how reliable the  data are, especially as the army is not an unbiased party to the conflict.  33  Table 3.  Summary of Impact Evaluation Results    Name of Project  Evaluation  Pov Reduction  Targeted toward poor,  Use/Access to Services  Governance  Social Capital Conflict  Years  (consumption/  vulnerable groups  (educ, health, watsan,    expenditure) roads)  SOUTH ASIA  1  AFGHANISTAN—  2007–09  X No  √ Yes for female groups in  √ Yes, water, electricity  √ Yes, village assembly  X No, trust,  X Limited  NSP2  Interim  terms of engagement in  X No transport, irrigation attendance, awareness  conflict,  effects  evaluation  community life, medical  Weak – health, school  of CDC issues, and  social  care, schooling and    more positive attitudes  cohesion  involvement in incomeâ€? toward govt, positive    gen. activities for women 2  INDIA—Andhra  2004–06   √ Yes  √ Yes, 15%  nutritional  N/A  Not measured  Not  Not  Pradesh DPIP  (2.5 years)  11% points,  gains, 50% increased  measured  measured  increase in assets  assets, not consumption  26% 3  NEPAL—Poverty  2007–10  √ Yes, 19% points  √ Yes  √ Yes, education  Not measured  X No  Not  Alleviation Fund  19% decline on  X No, health  measured    incidence of food  insecurity EAST ASIA  4  INDONESIA—  2003–07  √ Yes, 11% points  √ Yes  √ Yes, health  Measured in next  Measured in  ≈ Mixed  KDP2  for poor HHs  X No for femaleâ€?headed  X No, educ  survey  next survey    √ Yes, reduces  HHs  unemployment  1.5% 5  INDONESIA—   2006–07  √ Yes, 11%  √ Yes, conflict  √ Yes, economic uses  X No, trust in govt,  X No, trust,  X No  KDPâ€?BRA  reported by  communities,   and land  knowledge of govt  conflict,    village head,  ≈≈ At HH level conflict  X No, impact on educ,  affairs, attitudes about  social  increase in assets  and nonâ€?conflictâ€?affected  health, infra  govt.  cohesion  and farming of  persons benefited  land  X No impact on  employment  levels 6  INDONESIA—   2007–10  Not the objective  √ Yes, low health and  √ Yes, overall for educ &  Not measured  Not  Not  PNPM Generasi  education status areas.  health impacts  measured  measured  ≈ At HH level, only educ,  not health 34    Name of Project  Evaluation  Pov Reduction  Targeted toward poor,  Use/Access to Services  Governance  Social Capital Conflict  Years  (consumption/  vulnerable groups  (educ, health, watsan,    expenditure) roads)  7  INDONESIA—  2004–07  X No impact on  X No overall,   √ Yes, 3% for improved  X No, UPP is seldom  X No,  Not  Urban Poverty    per capita  but 9% improved access  access to sanitation (9%  mentioned as program  community  measured  Program II  consumption or  for sanitation among the  for poorest),  that deals with the  participation,  assets  poorest.  Project groups  X No. access to water  most important  membership  consisted mostly of the    problems of the urban  in  more educated, affluent,  village  organizations & officially connected  , and  contributions  8  PHILIPPINES  2003–06– √ Yes, 5% increase  √ Yes   √ Yes, health, secondary  √ Yes, attendance in  √ Yes, levels  ≈ Mixed,  Kalahiâ€?CIDSS  10  in per capita    school, college, more  village assemblies,  of trust,  decrease in  consumption  financing institutions  knowledge of village  membership  MILF areas,  X No, water and  issues, and attitude  in  increase in  elementary school  toward govt.  organizations  NPA areas  X No,    collective  action    AFRICA  9  SENEGAL PNIR  2002–04  √ Yes, 65%  √ Yes for poorer HHs, But  √ Yes, clean water,  Not measured  Not  Not  increase in HH  political patronage  health, child nutrition  measured  measured  exp.  related to which villages  receive funding  10  SIERRE LEONNE  2005–09  √ Yes  Not indicated in report  √ Yes, health,  X No, spillover on local  X No,  X No  —GoBifo  community centers,  governance  spillover  impact  schools, water, latrines  √ Yes, local govt actors  X No, impact  more active, improving  on crime &  community attitudes  violence  toward govt.        35    Name of Project  Evaluation  Pov Reduction  Targeted toward poor,  Use/Access to Services  Governance  Social Capital Conflict  Years  (consumption/  vulnerable groups  (educ, health, watsan,    expenditure) roads)  11  TANZANIA Social  2008    √ Yes, mildly proâ€?poor          Action Fund II  overall, progressive for  national geographic  targeting, w/in district  targeting neutral  12  ZAMBIA Social  1998–2000  Not the objective  √ Yes, rural  √ Yes, education, health  Not measured  √ Yes, rural  Not  Recovery Project  X No, urban  X No, urban  measured  II    LATIN AMERICA AND EUROPE & CENTRAL ASIA  13  BOLIVIA Social  1994–98  Not the objective  √ Yes, progressive for  √ Yes, health, watsan  Not measured  Not  Not  Investment Fund  poorest districts, poorer  X No, education  measured  measured  14  HONDURAS  1994–98  Not the objective  HHs benefited. Health &  √ Yes, education, health  Not measured  Not  Not  Social  education did better at  measured  measured  Investment Fund  reaching the poor.  15  NICARAQUA  1994–97  Not the objective  Sewerage fared worse.   √ Yes, education, health,  Not measured  Not  Not  Emergency  water  measured  measured  Social  Investment Fund  16  PERU Social  1994–97  Not the objective  √ Yes, education        Fund  (FONCODES)  17  ARMENIA Social  1998–99  Not the objective  ≈ Mixed  √ Yes, education & water  Not measured  ≈  Mixed  Not  Investment Fund  Regressive in rural,  X No, health  measured  progressive in urban    SUMMARY  Average #  Out of 9 projects,  Out of 16 projects, Out of 15 projects, Out of 5 projects,  Out of 8  Out of 5  of years  7 positive  generally positive to  generally positive  Positive to mixed   projects,   projects,  covered in  mixed  mixed to no  mixed to  IEs = 3.1  impact  negative   years        36  VII.  Cost Analysis    In  this  section,  we  examine  three  costâ€?related  issues.  The  first  is  the  per  capita  allocation  amounts  for  subgrants.  How  much  are  some  CDD  projects  allocating  for  block  grants?  This  issue  is  important  for  those  designing  new  CDD  programs  and  deciding  on  block  grant  allocations.        The  second  issue  is  community  contributions  and  a  comparison  of  how  the  different  CDD  programs  handle  those  contributions.  Lastly,  we  look at issues of cost effectiveness and rates of return using the CDD model.    1. Size of block grants/ per capita amounts   Several CDD programs allocate block grant amounts per village, municipality, or area.  The grant amounts are  determined  normally  by  several  criteria,  including  levels  of  poverty,  remoteness,  number  of  villages  per  municipality,  and/or  population.    The  costs  of  building  basic  infrastructure  (roads,  schools,  health  centers)  can be a factor as well.  Per capita amounts for subgrants and  microcredit activities ranged from $1.77 per  capita  per  annum  in  the  Philippines  to  approximately  $40  per  capita  ($200  per  family)  per  cycle  in  Afghanistan’s NSP (table 4).  Some of these amounts may have changed since the project phases quoted here.  Also, these calculations only pertain to community block grant amounts, not to technical assistance or other  administrative costs.    To  have  any  impact  on  income  or  nonâ€?income  poverty  levels,  it  is  important  to  ensure  that  there  is  an  adequate level of investment.  Many of the programs have recognized the need to provide more than oneâ€?off  grants.  The oneâ€?off nature of these programs generally does not lead to sustainable institutional reform or  social changes.  Several programs—such as Indonesia, India with revolving funds, the Philippines—provide  multiple rounds of support and interventions.  Table 4.  Block Grant Per Capita Allocation Amounts by Project    Program  Per capita allocation amount ($)  Allocation criteria  Afghanistan NSP  $200 per household  Per cycle  India AP  $17  $86 per family  $185 per incomeâ€?generating  Community organization submits a proposal for an  Nepal PAF   activity/HH (@$30 per capita)*  incomeâ€?generating activity for each HH in organization  $3–$6  Javaâ€?Bali, population and poverty, annual  Indonesia KDPII  $6–$6.70  Off Java, population and poverty, annual  Indonesia KDPâ€? $22.30–$27.10  By population and conflict intensity, one cycle  BRA  $1.77 to $29.10 (median of $9.40,  Philippines KC  By number of villages in municipality, annual  avge. $10.70)   Senegal PNIR  $4–$5  By population, annual  Sierra Leone  $16  One cycle  GoBiFo  Source:  PADs.   *Source: Parajuli et al. 2012.  37  2.  Community Contributions   Community contributions are often mandated by programs as a sign of community buyâ€?in to and ownership  of the program.  CDD programs handle community contributions in different ways. Some mandate a certain  percentage; others make it optional or use it as a consideration when awarding project grants.  Community  contributions can be difficult to measure and determine a specific value, as they are often calculated inâ€?kind  in  terms  of  community  labor,  materials,  or  land.    With  the  exception  of  the  Indonesia  and  Philippines  programs, the other programs require a community contribution of between 5 and 25 percent of the costs (on  average, 8 percent of the costs), with several programs waiving the contribution requirement in the case of  extreme  poverty  (Zambia  and  Bolivia,  for  example).    In  some  cases,  such  as  the  Armenia  Social  Fund  and  Senegal  PNIR,  project  stakeholders  lowered  the  contributions  part  way  through  the  program  because  the  level was considered too onerous.  But based on program completion reports, the majority of programs had  no  problem  meeting  or  exceeding  the  required  contribution  amounts.    In  the  case  of  Indonesia  and  the  Philippines, there is no mandatory amount required, but it is a consideration when deciding on funding.  The  table below shows the various ways the programs in this study handle community contributions.     Table 5. Levels of Community Contribution by Project  Community Contribution Project  Mandatory?  Initial Provision  Change/Results Afghanistan NSP  A minimum of 10% for the cost of  In NSP I and II, the community  Yes  infrastructure    contribution was 15 to 17% on average   India APDPIP  Between 5–10% of total subâ€?   Yes  project cost in cash or kind Nepal PAF  10% of the subproject   Community contribution was in general  Yes  much higher than 10%  Indonesia KDP2  No mandatory community  Community contributions generally  No  contribution required but a  averaged 17% of total subproject costs  consideration when judging  proposals  KDPâ€?BRA      No  KDPâ€?Generasi      No  Indonesia UPP2  Contributions encouraged, may    No  be in the form of labor,  materials, and/or land  Philippines   Similar to Indonesia’s KDP, no    No  KALAHIâ€?CIDSS  mandatory community        contribution required but a        consideration when judging      proposals  Senegal PNIR  15% in local investment fund  Following the project midâ€?term review,    component, in the form of labor,  some reductions in local financial  Yes  materials, and/or money.    contributions for specific projects such    as medium size water supply systems  took place, to reward social inclusion  (women’s involvement) and  maintenance of facilities already  funded    38  Community Contribution    Project  Mandatory?  Initial Provision  Change/Results  Sierra Leone  Communities were required to    Yes  GoBifo  make their own contributions in  the form of money, skilled or  unskilled labor, and/or materials  Percentage was not fixed.  Zambia Social  About 25% of the total spillover  Due to the inability to fund community  Yes  Recovery  costs, in cash or kind and  activities up to the level originally  Program II  ascertained by the field appraisal  anticipated, the amount of  beneficiary  team   contribution was reduced  Where the community is  obviously very poor and are  willing but unable to meet the  full contribution, the appraisal  team may recommend a lesser  contribution on a caseâ€?byâ€?case  basis  Bolivia  Communities participate    Yes  SIF2  normally with 5% of total  subproject cost and the  requesting agency with 15–20%;  also consideration of waving the  requirement for a few selected  very poor communities Honduras  Subproject beneficiaries would    Yes  SIF3  contribute in cash, labor,  materials, or land  Nicaragua  Community with 5% through    Yes  FISE2  money, kind, or labor Peru  On average, 10% of project    Yes  FONCODES  capital costs, adjusted to the  type of project, and poverty  levels. In addition, for certain  types of projects such as  irrigation, water and power  supply, communities would be  required to organize to collect  charges, to cover the recurrent  costs, and the municipality or the  relevant ministry would ensure  maintenance  Armenia  At least 15% of microâ€?project  Contribution was lowered from 15% to  Yes  SIF  cost; could be inâ€?kind or cash.  10% on average, with a 5% minimum  Contributions of food or goods  contribution from poor communities,  that can easily be monetized will  based on assessment of poverty and  also be accepted  capacity to pay  Source: World Bank project documents  39  3.  Cost comparisons/effectiveness  Several related studies show CDD’s cost effectiveness as compared to equivalent works built through other  government service delivery mechanisms.  The Philippines, Indonesia, and Nepal, for example, have between  13 to 39 percent lower costs depending on the type of investments. These cost savings come primarily from  8 eliminating  the  middle  man  or  contractor  overhead.   For  social  funds  (Bolivia,  Honduras,  Nicaragua,  etc.)  where the projects allowed greater community control over decisions and resources, unit costs were found to  be  lower  by  25  to  40  percent.  However,  when  the  projects  worked  through  private  contractors  and  government intermediaries, unit costs tended to be higher (Rawlings et al. 2003).  Other CDD/SF programs in  Africa—such as in Burkina Faso and Malawi—also report similar rates of cost savings (in the 15 to 40 percent  range) as compared to other programs producing similar quality investments.  It will be important that CDD  programs  run  regular  technical  audits  checking  quality  and  publicize  unit  cost  figures  to  ensure  cost  effectiveness and also minimize possible corruption.    4.  Rates of Return   Economic  rates  of  return  are  calculated  based  upon  numerous  assumptions  and  methodologies,  thus  they  should be considered with caution.  Rates of return for CDD infrastructure projects show an average of 18 to  53 percent returns.  On roads for example, benefits are seen through improved connectivity, savings in travel time, more frequent market trips, more motorized traffic, increased passenger movement, increased labor and agriculture machinery movement, and reduced unit costs of transport. Large benefits accrued from entirely new economic activities and access to markets that were made possible by transport infrastructure. For water projects, the returns come primarily from time savings. The India livelihood projects were showing internal rates of return for various livelihood activities of 40 percent and higher.                                                                      8  Projects  calculate  these  cost  savings  differently.  Some  projects  include  costs  of  technical/facilitation  assistance,  while  others  only  calculate the costs of constructing the infrastructure itself.  40  VIII.  Major Factors Affecting Implementation and Outcomes    The previous section described the impacts and results from various CDD impact evaluations.  In this section,  we  propose  a  few  hypotheses  describing  some  of  the  general  reasons  why  we  are  seeing  some  of  these  results.  What are some of the common design features that worked or failed and under what contexts?  How  do  we  maximize  welfare  outcomes?  As  we  did  not  randomize  these  various  hypotheses,  it  is  difficult  to  ascertain for certain the external validity of these theories.  However, to the extent that we were able to point  to some common features and perspectives derived from the evaluation reports, qualitative studies, project  implementation completion reports and discussions with some project and evaluation teams, we do so below.    Task teams and reports cite several common features for the generally positive impacts on poverty reduction,  access to and use of services, and geographical targeting:    Establishing a more participatory and inclusive model of service delivery despite a low starting point  of  financial  and  economic  crisis  and  conflict  in  many  of  the  countries.  Project  completion  reports  and  some  task  teams  point  to  the  core  participatory  design  element  of  the  CDD  approach.    First,  communities  participate  in  project  identification  and  in  some  projects,  and  participate  directly  in  the  decision  on  how  funds are used.  The subprojects meet communities’ selfâ€?identified needs, whether it is for roads or markets,  schools for their children, or microfinance for incomeâ€?generation activities. Direct community identification  of  their  development  needs  enhances  the  chances  of  their  actually  using  the  services.    The  participatory  model  opens  up  opportunities  for  communities  to  engage  in  the  development  process  and  benefit  from  improved services and infrastructure.     Providing high­quality and adequate facilitation and technical assistance.  The importance of having— and  retaining—highâ€?quality  project  and  technical  staff,  especially  on  the  implementing  agency  side,  was  mentioned consistently across many of the projects.  Wellâ€?qualified project staff with strong commitment was  seen as a key to making many of these programs successful and ensuring quality control. These staff include  skilled  engineers  to  help  with  standard  technical  designs  and  quality  supervision  of  construction;  line  department education and health staff who review the technical feasibility of proposals and provide villageâ€? level  assistance  to  ensure  improved  access  and  utilization  of  services;  and  facilitators  who  can  motivate  communities  and  engage  marginalized  groups.    Several  projects  also  mentioned  the  crucial  role  of  central  ministry project leaders and provincial/district staff who had a vision for the program and were committed  to participatory development. These actors can “make or breakâ€? a program.    The need for highâ€?quality staff becomes particularly acute in conflictâ€?affected situations where institutional  capacity  is  weak,  yet  there  is  a  political  imperative  to  deliver  goods  and  services  quickly,  such  as  in  Afghanistan.    In  some  cases,  they  rely  on  subcontracting  NGO  staff  to  carry  out  the  social  and  technical  facilitation functions, but in other countries such as Indonesia, regular project staff previously received some  training in conflict resolution work.    Capacity  building  for  communities  is  a  critical  element  to  obtaining  positive  results,  both  on  the  governance  side  as  well  as  communityâ€?management  of  resources.    Village  facilitators  play  a  crucial  role  in  mobilizing communities and marginalized groups, and raising the level of awareness of and participation by  communities  in  the  development  process.    Especially  in  India,  the  focus  on  project  facilitation  teams  and  capacity  building  for  local  community  groups  was  stressed.    Several  projects  also  cited  the  importance  of  41  training community groups on procurement, contracting, reporting, and/or business development as a means  of empowering communities and improving transparency and cost effectiveness.    Utilizing poverty maps to target resources to poor areas.  Most of the CDD programs used poverty maps  and  the  latest  statistical  data  on  household  welfare  to  determine  geographical  targeting.  As  a  result,  the  targeting for the most part favored poorer areas of the country.  As for household targeting, projects in India  and Nepal that supported more incomeâ€?generating activities did well.  In Nepal, they selected groups of poor  and  excluded  people  based  on  objective  criteria,  including  ethnicity,  caste,  gender,  and  household  food  sufficiency.  In  CDD  programs  that  funded  more  community  public  goods—such as  transport  infrastructure,  schools, and health centers—for the most part, the benefits reached the poorer quintiles, with some nonâ€?poor  benefiting as well.  In a few cases, such as Indonesia and the Philippines, the programs worked best in remote,  poorer areas where service access was limited; building a road, for example, opened up enormous economic  opportunities  and  access  to  markets,  schools,  and  health  services.    In  Indonesia,  households  in  less  poor  subdistricts saw either no benefit or negative impacts. As CDD programs scale up in geographical coverage to  possibly include richer areas, they must be careful to assess if investments remain proâ€?poor. One possibility is  to allow per capita funding envelopes to vary by poverty levels. Also, the use of objective criteria (e.g., ethnic  minorities,  persons  with  disabilities,  food  insecurity,  or  dropâ€?out  female  students)  could  be  explored  for  better targeting of specific marginalized and vulnerable subâ€?groups.    Block  grant  funds  are  used  for  economically  productive  purposes.    The  seven  CDD  programs  that  had  positive poverty impacts shared some commonalities in terms of block grant usage.  India’s Andhra Pradesh  DPIP,  Nepal’s PAF,  and  Indonesia’s  KDP/BRA  overwhelmingly  supported  primarily  microcredit  and  income  generation  activities.    Senegal’s  PNIR  and  Sierra  Leone’s  GoBiFo  also  supported  agricultural  infrastructure  activities and skills training. For public works, Indonesia’s KDP and Philippines’ KALAHIâ€?CIDSS program cited  the community’s choice of building rural roads to open up remote areas to markets, and of agricultural and  irrigation  facilities  to  allow  farmers  to  sell  more  of  their  products  and  improve  production  practices.    In  summary, the microcredit activities put financial resources in the hands of selfâ€?help groups to use for smallâ€? scale  income  generation  activities,  while  economic  infrastructure  public  goods  such  as  roads,  bridges,  and  some agricultural facilities provided employment and stimulated higher production and market activity.    Ensuring an adequate level of resources for investment.  As mentioned in the preceding section, for the  program to have any impact on income or nonâ€?income poverty levels, it is important to make sure that there  is  an  adequate  level  of  investment  over  a  period  of  time.    First,  the  block  grant  allocation  amount  to  a  community must take into account the cost of building basic infrastructure (e.g. building a road in a remote  area with few contractors will most likely cost much more than building infrastructure in a more periâ€?urban  area).    The  grant  amounts  are  also  determined  normally  by  several  criteria,  including  levels  of  poverty,  remoteness, population, as well as the presence of other programs. Second, many of the programs have also  recognized the need to provide more than oneâ€?off grants.  The oneâ€?off nature of these programs oftentimes  does not meet the multitude of needs in a poor community, nor does it lead to sustainable institutional reform  or  social  changes.    Several  programs—such  as  Indonesia,  India  with  revolving  funds,  and  the  Philippines— provide multiple rounds of support and interventions over years.    Growth in learning.  Encouraging flexibility in project design and implementation allowed national and local  governments to adjust practices over time and learn from experience with a new model.  Project stakeholders  speak  of  “growth  in  learning.â€?    Not  everything  can  be  done  at  the  same  time  or  from  day  one  of  project  effectiveness,  so  there  is  an  evolution  in  design  over  phases  and  even  within  the  project  lifetime  through  42  operational  manual  updates.  For  example,  several  countries  such  as  Indonesia  and  Bolivia  revised  their  operational  manuals during  the  life  of  the  project  to  improve  the  poverty  targeting  criteria or  to  add  more  social accountability measures. Timely studies and evaluations helped to inform program lessons, which were  then  assimilated  into  operations.    Several  programs  began  on  a  limited  scale  in  order  to  test  systems  and  build capacity first, before expanding during later phases to larger geographical areas.    Challenges in Achieving Results  As the illustrative CDD results chain in Figure 1 indicates, there are numerous challenges that governments  face in implementing these programs.  Many are not unique to CDD programs, but many factors do impact on  the programs’ abilities to reach their goals and targets.  Discussions with task team leaders and a review of  implementation  completion  reports  covering  the  respective  evaluation  period  reveal  the  following  issues  hindering the achievement of positive impacts:    Design Issues  Limited  impacts  on  governance  and  social  capital  spillovers.      This  study  showed  that  results  from  governance or social capital spillovers were mixed.  On the one hand, governments have expanded many of  these CDD programs nationally (Indonesia, Philippines, India, etc.) and put considerable domestic resources  into them.  So these CDD approaches are indeed expanding geographically in several country contexts.      However,  spillover  effects  outside  of  the  “project  bubbleâ€?  in  terms  of  the  way  government  officials  and  citizens’ approach and manage other development programs and civic activities have been very limited.  The  Philippines,  for  example,  showed  increased  levels  of  trust  within  project  communities  and  a  significant  increase in the proportion of respondents indicating that most people in their village can be trusted, as well  as a smaller but still positive impact on trust in local and national officials.  Spillover effects were seen in the  areas of  (a) increases in membership to organizations; (b) greater attendance in village assemblies; and (c)  more households who were aware of the income and expense details of their village local government unit.    However,  there  were  decreases  in  the  proportion  of  households  that  participated  in  communal  activities  (bayanihan), indicating some issues perhaps with opportunity costs.  In Sierra Leone, on the other hand, the  evaluation  revealed  no  treatment  effects  on  the  standard  proxy  measurements  for  social  capital—trust,  collective  action,  groups  and  networks,  inclusion,  and  information.    Related  to  local  governance  improvements,  however,  local  government  leaders  were  more  active  in  the  planning,  construction,  and  oversight of local public goods in treatment communities. The greater interaction appears to have increased  citizen confidence in their local representatives.      There could be several reasons for the general lack of spillover effects on social capital and governance. The  first  could  be  a  design  issue.  Many  projects  are  “ringâ€?fencedâ€?  and  project  actors  focus  their  efforts  only  on  making  their  operations  work.  Almost  all  the  social  funds  reviewed  for  this  study  started  as  “temporaryâ€?  programs with semiâ€?autonomous agencies, designed to deliver “oneâ€?offâ€? goods as an emergency response to  financial crises.  Thus, institutional sustainability and longerâ€?term reform were not really a focus in the earlier  program phases. 9 As many of the programs were started at a time of crisis, the fact that they could deliver  muchâ€?needed  services  and  assistance  at  all,  and  communities  were  satisfied  with  those  services,  may  lead  governments  to  continue  to  emphasize  service  delivery  only,  without  really  instigating  local  governance  reform.  In some cases—Indonesia, Afghanistan, and Zambia, for example—it could be that the government                                                                     9 In countries such as Bolivia, Honduras, and Peru, the social fund structures are still in place in one form or another, and are involved in  cofinancing of local government/municipal investments.   43  reform roadmap is not yet clearly defined or the decentralization agenda assigning roles and responsibilities  to local entities is ambiguous, and project designs reflect that lack of clarity and coherent vision. For various  social, political, and cultural reasons, behavioral norms encouraged or induced through the project may not  spill over into other spheres of community life.      The second reason could be due to a relatively short time span of measurement.  Among the five evaluations  that  did  try  to  measure  social  capital  in  a  rigorous  way,  the  measurements  occurred  after  one  year  of  implementation  (Indonesia’s  KDPâ€?BRA  showing  zero  effects)  to  seven  years  with  Philippines  Kalahiâ€?CIDSS  showing more positive effects and spillover impact.  Transformational societal and institutional change may  require longer periods of time to occur, and there could be a time lag in detecting these effects.      Third, given the amorphous, multiâ€?dimensional nature of some of these concepts of “empowerment,â€? “good  governance,â€? and “social capital,â€? one wonders whether the measurement tools in use are the correct ones or  if the proxy indicators are robust enough to capture these rich concepts, which are so heavily dependent on  social,  economic,  cultural,  and  political  context.    For  the  IEs  discussed  in  this  study,  a  mixture  of  tools  was  used for measurement of social capital: quantitative surveys, behavioral games, and qualitative techniques. It  would be worthwhile exploring further how these instruments can be more effectively applied.       Lastly, it is evident from the literature review that many factors contribute to governance and social capital  spillover  effects  such  as  reformâ€?minded  governments,  policy  and  regulatory  reforms,  predictability  of  and  repetition  of  activities,  and  more  open  and  conducive  environments  for  civic  engagement  and  openness.   Decentralization reforms in many contexts may not yet have created a more participatory environment at the  local level. For instance, what would be the incentive for a villager to participate in other local development  projects or planning if no funds existed through those channels or the local village chief had a monopoly on  those particular resources?  The demonstration effect of the CDD programs—with their participatory aspects  and effective delivery of services—may not prove strong enough to make a dent in decadesâ€?long institutional  norms and structures.  All these factors add to the mix of whether or not reform occurs.10   Understanding  better the transformational aspects of institutional and political change is a huge area of exploration on the  operational and research fronts.    Limitations  to  community  mobilization,  and  ability  to  mobilize  marginalized  groups.  Several  country  reports  cited  difficulties  engaging  specific  groups  of  communities,  e.g.,  seasonal  migrants,  refugees,  aged  persons, and destitute.  There was also the concern from programs that were scaling up regarding the ability  to cover the poorest of the poor or marginalized groups.  The active engagement of women also came up in  several cases as an area of concern.  This problem is related to the issue of household targeting, which has had  a  mixed  track  record  in  some  projects.    Several  of  the  CDD  programs  reviewed  in  this  study  had  positive  results in terms of reaching poor and marginalized groups—including Nepal, India, and the Philippines—and  it will be important to learn how those programs were able to do that operationally.    Implementation Challenges  Financial  disbursement  challenges.    Out  of  the  17  countries  covered  in  this  study,  nine  countries  mentioned  significant  problems  of  timely  and  adequate  disbursements  due  to  (a)  currency  devaluations  (especially  in  LAC);  (b)  difficulties  with  lateness  and  amount  of  government  counterpart  financing,                                                                     10 Comments based on discussions with several task team leaders and evaluators.  44  particularly at local levels; (c) late trust fund or donor financing; and (d) community counterpart financing.   These delays created difficulties in the predictability of funding.  Staff were able to maneuver around some of  these difficulties by reprogramming budgets, or in some cases cutting back some expenditure categories that  were not being used fully, such as capacity building or studies.    Conflict­affected  situations.    Seven  of  the  countries  covered  in  this  study—Armenia,  Nepal,  Afghanistan,  Sierra Leone, the Philippines, Indonesia and Peru—faced varying degrees of violent conflict, making program  delivery of resources and services difficult if not limited in certain areas.    Natural  disasters.    Five  countries—Senegal,  Armenia,  India,  the  Philippines,  and  Indonesia—experienced  setbacks due to natural disasters, including drought, tsunami, floods, earthquakes, and storms. This indicates  that CDD programs oftentimes must think of ways of using their networks and systems to respond to natural  disasters  and  build  community  resilience  in  times  of  emergency.    Several  programs  noted  that  the  onâ€?theâ€? ground presence of CDD programs and its project architecture allowed the programs to respond quickly and  flexibly to identify emergency local needs.  The programs served as a kind of “community safety net.â€?    Coordination  with  other  ministries.    Difficulties  in  coordinating  with  line  ministries  for  construction,  supervision, O&M, and technical quality support were mentioned in several projects.  As mentioned earlier,  technical  staff  are  crucial  to  assist  communities  and  enhance  access  and  use  of  services.    But  better  coordination with these line ministries would enhance impacts even further.    Operations and maintenance (O&M). Five programs mentioned the challenges ahead in maintaining locally  built  infrastructure  and  working  out  longâ€?term  arrangements  for  O&M.    Very  few  programs  indicated  satisfaction with sustainable O&M arrangements.    Monitoring and Evaluation. Several ICRs noted the challenges of putting a reliable management information  system in place, as well as the need for government counterparts to use the available information for adaptive  management and active monitoring.  Several task team leaders cited deficiencies in implementing agencies’  ability  to  analyze  information  from  process  monitoring  or  regular  project  reports  so  as  to  inform  decision  making.  Reports instead were used for public relations or to report upwards and to donors, rather than as  inputs  to  improve  project  management.  Information  from  impact  evaluations  tended  to  come  too  late  to  inform the next phase of design or scaleâ€?up.      What  the  review  makes  clear  is  that  there  is  not  one  linear,  straightforward  path  in  achieving  project  objectives or, for that matter, changing social or governance norms.  What is more evident when one looks at  some  of  the  factors  behind  why  the  programs  achieve  the  positive  or  negative  results  they  do,  is  that  undertaking institutional reforms is much more difficult than getting roads built or children immunized.  The  fact that many of these CDD programs were born as a direct response to overly centralized, inefficient state  bureaucracies means that changing the way governments and bureaucracies operate will take time and will  be a bumpy road.        45  IX.  Discussion of Operational and Research Implications  1.  Operational Implications of the Evidence  Several  of  the  programs  have  evolved  from  the  earlier  phases  covered  during  the  evaluation  period.   Nevertheless,  many  issues  remain  very  relevant  for  today’s  CDD  operations.  One  can  draw  several  operational implications from the evaluation findings and the discussion of implementation challenges:    ï‚· Use caution in determining objectives and what can realistically be achieved. In reviewing the  objectives and designs for these 17 CDD projects, the overwhelming focus was on delivery of services  through communityâ€?led,  participatory means.    The bulk  of the funds  (on average  80  percent)  went  toward  block  grants  or  community  investment  funds  for  infrastructure,  services,  and  incomeâ€? generating  activities.    On  the  poverty  reduction,  poverty  targeting  and  service  delivery  fronts,  the  evidence is generally positive and most projects met their objectives.  However on local governance  and social capital improvements, the findings are more mixed, with little evidence of spillover effects  outside  the  project  sphere.  For  future  CDD  designs,  project  stakeholders  may  wish  to  be  more  circumspect  and  realistic  about  how  ambitious  and  broad  the  objectives  should  be.    Poverty  reduction,  improved  use  and  access  of  services,  social  capital  improvements,  local  governance  reforms,  and  conflict  prevention  together  represent  by  any  measure  an  unrealistic  and  overly  ambitious  agenda,  especially  over  one  threeâ€?toâ€?four  year  project  phase.    Project  development  objectives should be more explicit and limited as to what they can realistically achieve.      ï‚· Local governance reforms and pathways of change.  In chapter VIII, we spoke about the positive  to  mixed  impacts  on  local  governance  in  the  programs  reviewed  for  this  study.    On  the  one  hand,  many  of  these  programs  were  scaling  up  nationally,  which  means  that  the  government  was  expanding the CDD approach nationwide. In some cases such as the Philippines and Sierra Leone, we  also saw evidence of governance spillover effects in the areas of (a) increases in attendance in village  assemblies; (b) more households who were aware of the finances of their village local government  unit; or (c) in the case of Sierra Leone, more local government leaders who were more active in the  planning, construction, and oversight of local public goods, leading to increased citizen confidence in  their local representatives. But in other programs, there seems to be more of a “project bubbleâ€? or  ringâ€?fence approach with little spillover into other governance norms outside of the project.  In the  future,  there  should  be  much  more  focus  on  identifying  the  pathways  for  governance  reforms,  and  the resources, enabling environment, and political will to achieve these changes sustainably.    ï‚· CDD Programs are beginning to have longer time frames.  The average duration of CDD projects  covered in this analysis is 11.8 years, ranging from Indonesia’s KDP/BRA which was only a oneâ€?year  operation  (as  a  response  to  an  immediate  postâ€?conflict  situation)  to  Bolivia,  Honduras,  and  Indonesia,  scheduled  to  run  approximately  16  to  17  years  at  this  point.    Some  of  these  programs  were preceded by similar programs—such as the Village Infrastructure projects in Indonesia, which  preceded KDP. India’s AP program built on UNDP’s program and earlier government investments in  women’s  selfâ€?help  groups.    In  many  cases,  this  means  these  programs  were  popular  enough  to  survive changes in political administration and government leadership.  And given these longer time  frames, there are opportunities to start thinking about the evolution of these programs over the long  term, what continuous investments could be made over time, what the phasing of reforms might be,  46  and what longâ€?term objectives are achievable.  The Bank should also put in place analytical work to  inform the design of future phases and impact evaluations to measure change over a longer period of  time.      ï‚· CDD and conflict­affected situations. This study includes four impact evaluations—for Afghanistan,  the Philippines, Indonesia, and Sierra Leone—that examined the programs’ impact on conflict.  Not  surprisingly,  there  is  no  impact  on  macro  levels  of  violent  conflict,  except  in  the  case  of  the  MILF  rebel  group  in  Mindanao,  Philippines.  These  findings  should  not  be  surprising,  given  that  many  of  these  violent  conflicts  are  endogenous  to  communities  and  beyond  their  control.    Nevertheless,  development programs in general—not only CDD interventions—must operate cautiously in conflict  and fragile situations as they can create severe harm or have positive impacts.  CDD programs can  attract  conflict  by  introducing  competition  for  funds,  bringing  more  development  funds  into  a  community,  exacerbating  existing  social  cleavages,  or  reinforcing  political  patronage  systems  of  largess.    On  the  other  hand,  if  designed  well,  they  can  also  address  longâ€?standing  grievances  of  exclusion  and  of  a  nonâ€?responsive  state,  e.g.,  MILF  in  the  Philippines,  or  introduce  community  mechanisms for mediating burgeoning conflicts (Indonesia).    ï‚· Investing  in  capacity  building  and  training  for  project  staff  and  community  groups.  The  projects  varied  widely  in  the  amount  of  emphasis,  time,  and  resources  they  placed  on  capacity  building  for  project  staff  as  well  as  community groups,  with  India placing  much more  emphasis  on  coaching,  mentoring,  and  capacity  building  of  both  project  staff  and  community  groups.    In  the  Philippines, the longer community facilitation process may have contributed to the larger effects on  social capital and local governance.  Interestingly, the Philippines evaluation observed improvements  in  village  governance  even  in  communities  that  were  not  prioritized  for  investments  and  had  only  received facilitation support (Labonne 2011).    ï‚· Operations and Maintenance of Investments, Sustainability. Arrangements for improved O&M of  subproject  investments  require  greater  attention  in  CDD  projects.    Whose  responsibility  is  it  to  maintain  the  infrastructure after  it  is  completed—the  communities, villages,  or  local  governments?   How  does  one  pay  for  recurrent  and  capital  costs  if  needed?    One  possibility  is  to  explore  piloting  various options for O&M arrangements within the CDD programs.      ï‚· CDD and social resilience.  As described in chapter IV, many of the CDD programs in this study grew  out of a situation of economic and financial crisis or conflict and fragility.  Therefore, CDD can be seen  as an important asset for the resilience of those societies in the face of future shocks.  Many of the  scaledâ€?up  CDD  programs  already  provide  the  “plumbingâ€?  and  community  mobilization  channels  needed  to  target  flexibly  those  who  fall  into  poverty  and  to  provide  a  rapid  means  of  delivering  resources  and  shortâ€?term  employment  to  communities.    The  ability  of  CDD  programs  to  build  community resilience and a communityâ€?led social safety net should be explored further in the future.    ï‚· What would CDD interventions look like in urban settings?  Global trends show that for the first  time  in  history,  more  than  half  the  world’s  people  live  in  cities,  with  an  estimated  1  billion  people  living in urban slums in developing countries. Over 90 percent  of urban growth is occurring in the  developing world, and the face of poverty is becoming urban.  Some of the earlier social funds—such  as  Bolivia,  Zambia,  and  Armenia—covered  both  urban  and  rural  areas  with  varying  degrees  of  success. The one urban CDD program included in this study—the Indonesia Urban Poverty Program  47  (UPP2)—had no impact on welfare outcomes, limited impact on access to services, and no impact on  social  capital  dimensions.    The  vast  majority  of  the  World  Bank’s  CDD  portfolio  operates  in  rural  areas,  and  most  of  the  projects  and  evaluations  covered  in  this  study  focused  primarily  on  rural  settings.    Providing  services  and  local  infrastructure  (roads,  bridges,  irrigation)  using  a  CDD  approach  may  be  appropriate  for  rural,  especially  remote settings,  but  may  not be  the  appropriate  intervention in urban areas where services and transport access are more readily available. Issues of  poverty  targeting,  a  menu  of  investment  options,  community  mobilization,  and  difficulties  working  with some highly transient populations will create challenges for the traditional CDD model in urban  settings in the future. Testing several urban CDD pilots in different regions would be advantageous.     ï‚· Working with supply­side actors. Related to local governance and reform issues, it is common for  projects  to  experience  difficulties  in  coordinating  with  various  line  ministries  for  coordinated  area  planning, service delivery, technical inputs, and O&M.  As several of the CDD national programs grow  and  expand,  the  agenda  in  this  area  needs  to  evolve  as  well.    How  can  CDD  programs  build  community  empowerment  to  hold  local  governments  and  technical  ministries  more  accountable?   Should  CDD  programs  start  providing  more  support  and  capacity  building  for  the  supplyâ€?side  of  solving problems of technical quality, e.g., teacher and health worker training or financial services?   Or should they bridge gaps between supply side and community groups, such as a community liaison  person within banks or technical agencies?  India and Indonesia provide a few examples in this area.   This topic opens up ample opportunities for multiâ€?country operational pilots and research.     2.  Implications for Monitoring and Evaluation  ï‚· Management  information  systems  and  reporting.  Several  reports  mentioned  the  difficulties  of  putting  management  information  systems  in  place  and  capturing  basic  project  information  for  thousands  of  transactions.    Task  teams  also  raised  the  challenge  for  government  implementing  agencies of  feeding  reliable,  timely  monitoring  information  into analysis  and  management  decision  making. A great deal of technical assistance is needed in this area.      ï‚· Impact  Evaluations.  On  the  evaluation  side,  on  the  positive  front  there  are  more  rigorous  impact  evaluations  in  place  than  there  were  10  years  ago.  However,  even  these  are  miniscule  in  number  despite 400 active CDD programs and 25 years of implementation.  The fact that there are only 17 or  so  rigorous  impact  evaluations  available  at  this  point  means  that  more  strategic  attention  and  resources  need  to  be  placed  in  this  area.    As  of  November  2011,  the  WB  official  impact  evaluation  database  had  approximately  50  CDD/SF  evaluations  listed;  however,  only  these  17  had  been  completed, with others aborted and some in progress. In order to build a better evidence base, it will  be  important to ramp  up  the  number  of  strategic  impact  evaluations  and  also  ensure  that they are  geared  toward  answering  longerâ€?term  sustainability  issues,  as  well  as  those  areas  where  little  evidence is available (see section below).    Second, impact evaluations take an enormous amount of time, resources, and specialized expertise.  The  technical  challenges  of  carrying  out  robust  evaluations  are  formidable,  especially  in  countries  where specialized expertise is scarce. This study looked at several evaluations that were aborted or  could not be included in this analysis because of poor design, e.g., only beforeâ€?after comparisons with  no  control  groups,  or  inadequate  staffing  to  develop  proper  sampling  frames  and  high  quality  questionnaires, or undertake inâ€?depth data analysis. Many governmentâ€?procured evaluations are not  48  of  high  enough  quality  because  of  poor  design  issues,  insufficient  staffing,  and  lack  of  rigorous  technical review. For many of the studies covered under this review, task teams have come to rely on  trust  fund  resources  to  fund  independent  studies  and  evaluations,  and  to  bring  in  much  needed  technical advice on evaluation design, sampling frames, and analysis.  There is a move now within the  WB  to  undertake  more  strategic  impact  evaluations  as  well  as  raise  resources  for  these  important  evaluations  but  as  yet,  sufficient  resources  have  not  yet  materialized.  As  impact  evaluations  are  a  global  public  good,  we  would  support  greater  efforts  by  the  WB  and  other  donors  to  set  aside  separate funds and resources for undertaking strategic impact evaluations on CDD.    Lastly, the timing of impact evaluations is crucial. They must be designed at the outset of the projects  as part of geographical area selection and to establish baselines of treatment and control.  Experience  also  indicates  that  it  has  been  very  difficult  to  complete  impact  evaluations  in  a  timely  manner  to  inform  decision  making  on  future  phases  and  possible  scaleâ€?up.    The  evaluation  community  must  make  better  efforts  to  plan  evaluations  in  such  a  way  that  results  can  feed  into  timely  decision  making  about  future  project  design.    Finally,  evaluations  measuring  welfare  and  social  capital/governance changes over time will be important.  Based on this study, the average amount of  project  intervention  time  measured  by  the  impact  evaluations  was  relatively  short,  only  3.1  years.  Even within those years, the number of cycles or “treatment interventionsâ€? varied significantly. Some  areas  could  have  received  only  one  round  of  treatment  during  the  period  or  annual  treatments.   Evaluations over a longer period of time are needed to assess longerâ€?term impacts.       ï‚· Quality control of evaluations ­ The last issue worth highlighting is one of quality control and peer  review. This study left out several studies because the methodologies were not of sufficient rigor or  evaluations  were  not  peer  reviewed.  Another  measure  to  enhance  evaluation  quality  would  be  the  regular use of preâ€?analysis plans for evaluations to limit issues of “data miningâ€? and “cherry pickingâ€?  of  results.    While  they  do  not  solve  all  issues  of  quality  control,  these  steps  to  undertake  a  preâ€? analysis  plan  and  ensure  quality  review  of  concepts,  evaluation  methodologies,  and  final  reports  would be a move in the right direction to enhance the quality and rigor of the studies.      3.   Implications for Further Research: Remaining Unanswered Questions  The  literature  review  opens  up  several  lines  of  useful  inquiry.  Earlier,  we  had  mentioned  the  need  to  undertake  further  research  on  the  lack  of  social  capital/governance  spillover,  as  well  as  analytical  work  to  inform the next project phase.  In addition, a few more topics require exploration:    ï‚· Governance reform and building a generation of reform­minded local leaders. We have already  identified the need to build more evidence on local governance  spillovers.  One unexplored area of  study  is  to  track  systematically  CDD  project  actors  over  time  to  see  if  they  assume  leadership  positions  in  government  or  NGOs.    There  are  ample  anecdotal  accounts  of  community  facilitators  being elected to local village councils or regional parliaments, but not much by way of documented  evidence.  Many of the largeâ€?scale CDD programs now work with and train tens of thousands of local  facilitators,  project  implementers,  and  local  government  officials.  Undertaking  longitudinal  tracer  analyses of these local actors will be important to understand if CDD is impacting on building a new  generation of national development leaders.    49  ï‚· Comparing  CDD  approaches  versus  non­CDD  approaches  using  other  government  service  delivery mechanisms. As mentioned earlier, none of the impact evaluations in this study compare  the  outcomes  of  CDD  investments  with  those  of  alternative  interventions  such  as  direct  central  or  local  government  service  delivery.    Some  cost  comparisons  are  made  in  chapter  VII,  but  not  for  outcomes.  Part of the reason for this stems from the fact that these CDD/social fund programs grew  as a response to the failure of decades of topâ€?down, centrally driven service delivery, so governments  do  not  feel  they  need  to  experimentally  test  this  hypothesis.  Nevertheless,  running  a  randomized  experiment  comparing  these  different  approaches  along  the  impact  dimensions  discussed  in  this  paper would be valuable.    ï‚· Reaching  marginalized  groups.  Some  programs  were  more  successful  than  others  at  reaching  marginalized  and  excluded  groups.  Social  inclusion  is  becoming  a  major  policy  issue.  Within  countries, lagging regions, growing income inequalities, and exclusion of vulnerable groups are some  of the major challenges for government policy makers in developed and developing countries alike.  Several  of  the  CDD  programs—such  as  Nepal,  India—are  worth  looking  at  further  for  comparative  work on improving household targeting, types of information, outreach, and capacity building efforts,  and other implementation modalities for reaching these groups.  Again, ethnographic work and other  qualitative  studies  that  allow  for  “saturation  of  informationâ€?  in  these  specific  areas  could  be  very  enlightening.    ï‚· Unpacking  the  black  box  of  decision  making.    While  the  impact  evaluations  and  final  reports  focused on outcomes and outputs—roads built, school enrollments, etc.—there was little discussion  or  documentation  about  how  these  decisions  were  made  at  the  community  and  local  government  levels.  Several  studies  pointed  to  community  satisfaction  with  the  program  and  its  services  and  whether  or  not  the  subprojects  matched  community  and  local  officials’  preferences,  but  more  qualitative  work  should  be  done  to  understand  the  decisionâ€?making  process  in  the  allocation  of  resources.  Is there full, informed participation regarding these decisions, or are the projects merely  reinforcing patronage systems and political patterns of largesse? Why are some subprojects chosen  and  not  others?  How  do  we  ensure  that  decisionâ€?making  bodies  remain  accountable  and  are  not  captured by preâ€?existing local elites?  How do traditional power holders in the community—such as  the  village  chief  or  tribal  leaders—view  these  CDD  programs  and  their  mechanisms  for  decision  making?  How  are  marginalized  groups—ethnic  minorities,  persons  with  disabilities,  etc.—involved  in the decision making, if at all?  Some more qualitative and ethnographic work in this area would be  useful to complement the quantitative evaluations.    ï‚· Sustainability. Issues of sustainability came up repeatedly in the document review.  There are many  facets to sustainability in the CDD context. First, there is the issue of infrastructure and investment  sustainability;  that  is,  sustainability  regarding  CDDâ€?produced  infrastructure  and  how  to  maintain  those  works.    More  longtitudinal  technical  quality  studies  looking  at  the  sustainability  of  the  community infrastructure or the feasibility of the incomeâ€?generating activities would be informative  for  future  design.  Highâ€?quality  design  and  sound  construction  are  major  factors  in  determining  sustainability.    The second issue concerns examining the sustainability of institutions.   Sustainability also was an  issue mentioned in the context of social funds. Unlike some of the later CDD programs, many social  funds  were  designed  originally  as  transitional  institutions  to  operate  efficiently  and  transparently  50  until the institutional capacity of state social sector entities were strengthened. The last 15 years or  so  has  seen  greater  attention  placed  on  the  longâ€?run  impacts  of  community  investments  and  institutional capacity (Van Domelen 2003).      Sustainability  of  impacts.  As  mentioned  earlier,  it  will  be  important  to  have  more  evaluations  looking  at  program  impacts  over  time.    The  longest  time  span  we  have  in  this  study—  the  KALAHIâ€?CIDSS  program— measures  over  7  to  8  years.    Positive  impacts  may  attenuate  as  initial  high  returns  from  low  baseline  situations  do  not  continue  and  investments  yield  diminishing  returns.  On  the  other  hand,  positive  impacts  could also increase as project actors become more used to the processes and repeated treatments add greater  expertise in planning and managing resources.  Some longer term evaluations will shed light in this regard.    As many of these programs continue, it will be important to measure whether program impacts attenuate or  change.    As  mentioned  above,  impact  evaluations  covering  longer  time  periods  will  also  be  important  to  assess the sustainability of welfare, social capital, and governance impacts.      51  X.  Concluding Remarks  This  study  has  focused  on  the  available  rigorous  evidence  to  date  across  17  WB  CDD  programs  and  the  operational and research implications of the findings. It highlights the historical evolution of the World Bank’s  CDD  portfolio.    The  vast  majority  of  the  CDD  programs  examined  in  this  study  grew  out  a  situation  of  economic, financial, or political crises. The social funds in Latin America, begun in the late 1980s and 1990s,  were  a  direct response to austerity  measures  put  in  place  under structural  adjustment  programs.    Many of  these  programs  also  tried  to  introduce  a  fundamentally  new  model  of  development  that  went  against  the  traditional  model  of  topâ€?down,  centrally  driven  development.  In  this  new  CDD  model,  communities  would  lead and participate in the planning, decision making, and management of local public resources to meet their  selfâ€?identified needs.      For  many  of  the  longer  running  CDD  programs,  these  initial  phases  of  reacting  to  crises  and  establishing  a  participatory model of development have passed.  The duration of the CDD programs covered in this study  averages 11.8 years, with a range of 1 to 14 years.  Positive achievements have been made in general on the  poverty reduction front, increased access and use of services and geographical poverty targeting, with more  mixed  results  on  the  household  level  of  targeting.    Evidence  on  the  social  capital,  local  governance,  and  conflict  resolution  areas  is  more  mixed  and  limited.    Depending  on  the  country  context,  the  overwhelming  emphasis by governments on socioeconomic welfare goals and service delivery through participatory means  is  likely  to  continue.    This  may  especially  be  true  in  fragile  and  conflictâ€?affected  countries,  as  well  as  those  countries still tackling high levels of vulnerability and pockets of poverty.  However, for this next generation  of longer running CDD programs, numerous issues emerge, such as continuing poverty reduction and service  delivery  efforts  in  an  effective  and  sustained  manner;  reaching  effectively  marginalized  groups;  addressing  pockets  of  poverty  and  lagging  regions  in  middleâ€?income  settings;  institutional  reform  from  the  bottomâ€?up  and  community  participation  in  governance;  building  inâ€?country  capacity;  enhancing  community  resilience  and social safety nets; and urbanization.    This  longer  term  agenda  has  numerous  implications  on  the  operational  and  research  fronts.        On  the  operational front, these include (a) clearly defined, realistic objectives on how much can be achieved on the  governance/reform  front  within  the  political  context  and  resources  available;  (b)  with  the  government,  developing clear roadmaps for institutional and governance reform; (c) poverty targeting issues, especially at  the community and household levels; (d) examining what CDD project design adaptations are needed in an  urban  setting;  (e)  understanding  better  the  impacts—both  positive  and  negative—of  CDD  interventions  in  fragile and conflictâ€?affected areas and approaching these situations cautiously in design; and (f) developing  stronger modes of collaboration with supplyâ€?side actors.    On the monitoring and evaluation front, this study points to the need for more highâ€?quality evaluations that  are  timely  and  in  some  cases  longer  term.  On  the  monitoring front,  the  need  for more  effective  monitoring  and reporting systems to provide information in real time was identified as a critical area for adaptive and  risk  management  and  informing  decisions.    One  of  the  cited  strengths  of  this  model  has  been  the  ability  to  adapt, with “growth in learning.â€?  Timely monitoring and supervision information helps enormously to make  this learning and adaptability possible.     Finally,  this  study  identifies  several  vital  research  opportunities  to  inform  the  CDD  portfolio.  There  is  certainly a growing need for more longitudinal studies examining longer term impacts of the program and the  52  sustainability of CDD investments.  Social inclusion also is a prominent and growing issue around the world,  both in terms of lagging regions as well as the exclusion of various identity groups and populations.  Several  country cases mentioned in this report provide successful examples of reaching these groups, but much more  comparative analytical work—and operational pilots— are needed to shed light on how to more effectively  support these  groups  in other  country contexts.   Lastly,  the  CDD  community  would  benefit  from  additional  qualitative and ethnographic work on how decisions are made at the local level.          53  ANNEX A.  Summary of Projects and Impact Evaluations      PROGRAM INFORMATION  IMPACT EVALUATION INFORMATION    Name of CDD  Total Program  Project Development  Evaluation  Period  Evaluation  Program  Period*  Objective **  Authors  Covered  Methodology  (Program  During IE  phase relevant  for  IE)  SOUTH ASIA  1  Afghanistan:  2003 to 2015  (a) to lay the foundations for a  Beath,  Christia,  2007–09 (2  Randomized  National  (2006 to 2011)  strengthening of community  Enikolopov,  years)  design, panel  Solidarity  level governance, and (b) to  Kabuli, 2010  survey in 500  Program II  support communityâ€?managed  villages in 10  subprojects comprising  districts, 6  reconstruction and  provinces;   development that improve  15,000  access of rural communities to  respondents, with  social and productive  FGDs   infrastructure and services    2  India: Andhra  2000 to 2016  To improve opportunities for  Deininger,  Liu,  2004–06  Propensity score  Pradesh  (2000 to 2006)  the rural poor to meet priority  2009  (2.5 years  matching, double  District  social and economic needs in  exposure)  difference,  2,400  Poverty  the six poorest districts of  household panel,  Initiatives  Andhra Pradesh, namely  51 counties in 3  Program  Chittoor, Srikakulam,  districts   (DPIP)  Adilabad, Vizianagaram,  Mahabubnagar, and  Anantapur.  3  Nepal:  2004 to 2014  To improve living conditions,  Parajuli,  2007–10  Randomized  Poverty  (2008 to 2014)  livelihoods and empowerment  Acharya,  phaseâ€?in  Alleviation  among the rural poor, with  Chaudhury,  approach, 200  Fund   particular attention to groups  Thapa, 2012  villages, 3,000  that have traditionally been  HHs sampled  excluded by reasons of  gender, ethnicity, caste and  location    EAST ASIA  4  Indonesia:   1998 to 2014  To (a)  support participatory  Voss, 2008  2003–07  Propensity score  Kecamatan  (2002 to 2007)  planning and development  matching, double  Development  management in villages; (b) to  difference,  Program II  support a broad construction  w/qualitative  program of social and  study  economic infrastructure in  poor villages; and (c) to  strengthen local formal and  informal institutions by  making them more inclusive,  accountable, and effective at  meeting villagers’ selfâ€? identified development needs  5  Indonesia:  1998 to 2014  To obtain 12 education and  Olken,  Onishi,  2007–10  Randomized   PNPMâ€?Rural  (2007 to 2010)  health targets related to  Wong, 2011  design across five  Generasi  improved school enrollment  provinces, over  Sehat dan  and attendance and child and  45,000  Cerdas (PNPM  maternal health  respondents,  Generasi)   w/ qual study  54      PROGRAM INFORMATION  IMPACT EVALUATION INFORMATION    Name of CDD  Total Program  Project Development  Evaluation  Period  Evaluation  Program  Period*  Objective **  Authors  Covered  Methodology  (Program  During IE  phase relevant  for  IE)  6  Indonesia:   2006 to 2007  To assist conflictâ€?affected  Barron,  2008–09  Propensity score  Kecamatan  communities in improving  Humphreys,  matching  Development  their living conditions through  Paler,  combined with an  Program/ BRA  provision of small projects  Weinstein,  instrumental  that accord with their needs  2009  variables  approach to deal  with issues of  noncompliance  with treatment  assignment,  qualitative study  7  Indonesia:  1999 to date  (a) to establish or support  Pradhan,  Rao  2004–07  Propensity score  Urban Poverty  representative and  Rosemberg.  matching, double  Program II  accountable community  2009  difference  organizations that are able to  estimates, 230  provide services to the urban  urban villages,  poor and increase the voice of  18,000  the poor in public decision  respondents  making; (b) to make local  government more responsive  to the needs of the poor  through increased  cooperation with community  organizations; and (c) to  improve services for the  urban poor (financial services,  social services, and  infrastructure)  8  Philippines:   2002 to 2014  To assist the borrower in  Asia  Pacific  2003, 2006,  Propensity score  Kapitâ€?bisig    strengthening local  Policy  Center,  2010  matching, double  Laban Sa  communities' participation in  May  2011,  difference, w/  Kahirapan –  village govemance, and  Labonne, 2011  qualitative study  Comprehensiv developing their capacity to    e and  design, implement, and    Integrated  manage development    Delivery of  activities that reduce poverty    Social Services    Program    (KALAHI    CIDSS)                             55      PROGRAM INFORMATION  IMPACT EVALUATION INFORMATION    Name of CDD  Total Program  Project Development  Evaluation  Period  Evaluation  Program  Period*  Objective **  Authors  Covered  Methodology  (Program  During IE  phase relevant  for  IE)  AFRICA  9  Senegal:  2000 to 2012  To (a) improve local  Arcand  and  2002–04  Propensity score  Programme  (PNIR/PLDP)  governance and local  Bassole, 2007  matching  National  (2000 to 2005)  capacity; (b) establish  d’Infrastructur participatory and  es Rurales  decentralized mechanisms for  (PNIR)  selecting, funding and  implementing rural  community investment  programs; (c) strengthen the  national institutions  supporting decentralization;  and (d) implement basic  infrastructure in a selected  number of rural communities.  10  Sierra Leone  2005 to 2010  (a) to strengthen social capital  Casey,  2005–09  Randomized  GoBifo Project  (2005 to 2009)  by enhancing the capacity of  Glennerster,  experimental  (JSDF)  villages and local  Miguel, 2011  design, 236  governments to design and  villages and 2,832  implement strategic  HHs;  focus group  development plans at village  discussions;  and ward levels; and   structured  (b) to enhance the  communities  accountability of decision  activities  makers through mechanisms  that foster open and  transparent governance    11  Zambia Social  1991 to 2005  To assist the government of  Chase  &  1998–2000  Propensity score  Recovery  (1995 to 2000)  Zambia’s poverty reduction  Sherburneâ€? matching, and  Project II  program through financing of  Benz, 2001  pipeline matched  community initiatives and  comparisons  building of capacity for    poverty analysis and    monitoring                                        56      PROGRAM INFORMATION  IMPACT EVALUATION INFORMATION    Name of CDD  Total Program  Project Development  Evaluation  Period  Evaluation  Program  Period*  Objective **  Authors  Covered  Methodology  (Program  During IE  phase relevant  for  IE)  LATIN AMERICA AND CARRIBEAN  12  Bolivia Social  1987 to 2004  To continue assisting the  Newman,  1994–98  Propensity score  Investment  (1993 to 1998)   government to improve  Pradhan,  matching,  Fund II  coverage and quality of  Rawlings,  matched  health, water supply,  Ridder,  Coa,  comparison; for  sanitation and education  Evia, 2002  education  services targeting primarily  evaluation,  the rural poor, including  randomized  indigenous peoples,  design.  disadvantaged women, and    children  13  Honduras  1990 to 2012  (a) to continue to assist the  ESA  1994–98  Pipeline matching  Social  (1995 to 1999)  government in strengthening  Consultores,  Investment  its poverty alleviation efforts  2005,  World  Fund III  and in maintaining social  Bank 2003  cohesion, while the  government attempted to  regain fiscal balance and the  line ministries strengthened  their institutional capacity;  and (b) to support the  government's decentralization  strategy, development of the  local contracting industry, the  sustainability of subprojects,  and the targeting of scarce  resources to the poorer  communities   14  Nicaragua  1992 to 2006  To help the government  World  Bank,  1994–97  Propensity score  Emergency  (1994 to 1997)  sustain its poverty alleviation  2000  matching;  Social  efforts and maintain social  oversampling  Investment  cohesion during the economic  1998 LSMS  Fund (FISE) II  adjustment period;   survey, 1998–99  additionally, the project  FISE exâ€?post  would help support the  survey  government's decentralization    strategy, strengthen the local    contracting industry, broaden    the menu of projects financed    by FISE, support the targeting    of resources to more    depressed areas, and help    meet the special needs of the    indigenous population                    57      PROGRAM INFORMATION  IMPACT EVALUATION INFORMATION    Name of CDD  Total Program  Project Development  Evaluation  Period  Evaluation  Program  Period*  Objective **  Authors  Covered  Methodology  (Program  During IE  phase relevant  for  IE)  15  Peru Social  1993 to 2000  To assist the borrower in (a)  Paxon  and  1994–97  Matched  Fund  (1993 to 1997)  sustaining its poverty  Schady,  2002,  comparison,  (FONCODES)  alleviation efforts, (b)  Instituto  oversampling  mitigating the social costs of  Apoyo, 2000  LSMS surveys,  its macroeconomic  National  adjustment, and (c)  Statistical  strengthening the institutional  Institute HH  capabilities of FONCODES    survey       EASTERN AND CENTRAL ASIA  16  Armenia  1996 to 2012  To support the lower income  Chase, 2002  July 1, 1998  Propensity  score  Social  (1996 to 2000)  groups among the Armenian  to June 30,  matching,   Investment  population through  1999 (HHs  pipeline matching  Fund (ASIF)  improvement of basic social  to report  services and creation of  changes in  employment opportunities.   previous 5  The primary objectives of the  years)  project would be to (a)  rehabilitate basic smallâ€?scale  infrastructure that can result  in immediate improvements  in the living conditions of the  poorest among the  population; (b) strengthen the  capacity of private smallâ€?scale  contractors and other microâ€? businesses so they can benefit  directly or indirectly from  contracts financed by the  ASIF; (c) generate  employment through  financing of labor intensive  public works; and (d) build  greater capacity for policy  makers to monitor and  analyze trends in the level and  structure of poverty in  Armenia    SUMMARY  Average      Average    duration of  number of  projects = 11.8  years  years  covered in  IEs = 3.1  years    Notes: *Total Program Period is the entire period of the program including several phases of the same program. **PDO is the  PDO listed in the project appraisal document covering the period for which the impact evaluation is relevant. For projects with  various phases, the PDO may change as the project progresses.        58  ANNEX B.  Individual CDD Project Descriptions     AFGHANISTAN: NATIONAL SOLIDARITY PROGRAM II  Total Project Period: 2002–15 (Project phase covered by IE: 2006–11)  Total amount for this phase: $120 million (IDA $120 million) with an additional funding of $75 million.     Background   More  than  two  decades  of  conflict  had  resulted  in  extensive  destruction  of  infrastructure  and  massive  population  displacements.  A  severe  drought  lasting  from  1999  to  2002  further  increased  rural  poverty,  depleted available assets, and led to further displacements.  The country has not fully emerged from its state  of  conflict  and  the  Taliban  and  other  insurgent/antiâ€?government  elements  continue  to  be  a  threat.  Opium  production  and  poppy  cultivation  continue  to  flourish  in  certain  parts  of  the  country  despite  government  attempts  at  eradication  and  alternative  livelihood  campaigns.  Afghanistan  is  ranked  the  second  poorest  country  in  the  world,  with  70  percent  of  the  population  living  below  the  poverty  line  of  $2/day.  The  vast  majority  of  the  population—about  80  percent—live  in  rural  areas,  where  the  poverty  rate  is  even  higher,  literacy  rates are  lower, and  services  are  more scarce.  Compounding  the  problem is  that  institutions  of the  government are weak and service delivery below the provincial level is limited.  The government introduced  the  National  Solidarity  Program  in  2003  as  a  followâ€?up  to  the  communityâ€?driven  development  (CDD)  component of the Emergency Community Empowerment & Public Works Project begun in June 2002.     Project  Development  Objectives  were:  (a)  to  lay  the  foundations  for  a  strengthening  of  communityâ€?level  governance,  and  (b)  to  support  communityâ€?managed  subprojects  comprising  reconstruction  and  development that improve access of rural communities to social and productive infrastructure and services.      Project  Components  were:  (a)  block  grants  for  communities  ($  94.3  million)  to  carry  out  subprojects  involving reconstruction and development activities through a facilitated participatory planning process; (b)  community  facilitation  and  subproject  preparation  ($9.7  million)  to  support  local  communities  through  facilitation  exercises  and  technical  assistance  to  establish  inclusive  community  institutions,  identify  local  development  needs  and  priorities,  and  prepare  subproject  proposals;  (c)  CDC  capacity  development  ($14.4  million); (d) implementation and management support ($1.5 million).     Block Grant Size   Block  grants  were  allocated  to  communities  at  a  rate  of  $200  per  family  with  an  upper  limit  of  $60,000  (corresponding to the construction cost of a quality primary school) per community.  With an average village  comprising  around  150  families  (equivalent  to  900  persons),  the  average  block  grant  was  approximately  $26,000. In phase II, there was an increase in the average amount of block grant per community to $33,500,  due to the return of internally displaced persons and refugees since 2003.      Types of Subprojects  Communities  can  utilize  their  block  grant  to  finance  projects  prioritized  in  their  community  development  plan.    In  NSP  I  and  II,  communities  were  given  a  choice  of  financing  infrastructure  or  human  capital  development (HCD) projects. Communities primarily prioritized infrastructure projects (80 percent), which  included water supply and sanitation (24 percent), rural roads (24 percent), irrigation (17 percent), village  electrification (14 percent), and human capital development (12 percent).   59  INDIA: ANDHRA PRADESH DISTRICT POVERTY INITIATIVE PROGRAM  Total Project Period: 2000–16 11(Project phase covered by IE: 2000–06)  Total amount for this phase: $118.2 million (IDA $111 million)     Background   In Andhra Pradesh (AP), despite concerted efforts to reform public sector antiâ€?poverty programs nationwide,  there was a progressive crowding out of development expenditures by costly and poorly targeted subsidies, a  rapidly expanding civil service, and increasing interest payments.  Of particular concern were the poor and  vulnerable households in regions dependent on riskâ€?prone rainfed agriculture, as well as those who lacked  productive assets and skills or those who suffered from ill health, disability, or illiteracy. To meet the needs of  these  poor  and  vulnerable,  the  government  of  AP  recognized  the  need  for  new  policies,  institutions,  and  processes, linked to direct assistance, designed to empower poor rural communities in the targeting and use  of resources. The Andhra Pradesh District Poverty Initiatives Project (APDPIP) was launched, building on the  United Nations Development Program (UNDP)â€?supported South Asia Poverty Alleviation Project (SAPAP) and  the AP government’s investments in institutionâ€?building through women’s selfâ€?help groups (SHGs).      Project Development Objective. The project objective  was to improve opportunities for the rural poor to  meet  priority  social  and  economic  needs  in  the  six  poorest  districts  of  Andhra  Pradesh,  namely  Chittoor,  Srikakulam, Adilabad, Vizianagaram, Mahabubnagar, and Anantapur.      Project Components were: (a) institutional and human capacity building ($20.34 million), which sought to  develop  selfâ€?reliant  and  selfâ€?managed  communityâ€?based  organizations;  (b)  a  community  investment  fund  (CIF) ($76.39 million) that would finance demandâ€?driven subprojects to be proposed by CIGs/SHGs, either on  their own or jointly with communities during implementation; (c) educational support for girl child laborers  and school dropâ€?outs ($30.42 million), which was intended to address the twin problems of child labor and  the high rate of school dropâ€?outs and finance a package of interâ€?related activities involving both nonâ€?formal  and formal education; and (d) project management, monitoring and evaluation, and studies ($7.65 million).     Block  Grant  Size.  The  project  was  successful  in  facilitating  implementation  of  36,477  income  generating  subprojects/  village  organization  level  micro  plans  involving  utilization  of  about  $75  million  (Rs.  3,288  million)  of  community  investment  funds  during  the  project  period.    The  project  investment  was  approximately $86 per household (or $17 per capita).     Types of Subprojects. The establishment of the CIF and the financing of subprojects/micro plans of the poor  have contributed to the diversification of the income generating activities pursued by the members. A large  proportion of the members preferred to invest in household dairy (30%), agriculture (29%), nonâ€?farm trade  (20%), and sheep rearing (10%), accounting for the bulk of investments.                                                                         11 APDPIP was followed by the Andhra Pradesh Rural Poverty Reduction Project (2003–11). The district-level project concluded in September 2011; a national-level follow-up CDD program—“National Rural Livelihood Missionâ€?—will be operating from 2011 to 2016. 60  NEPAL: POVERTY ALLEVIATION FUND PROJECT II  Total Project Period: 2004–14 (Project phase covered by IE: 2008–14)  Total amount for this phase: $109 million (IDA $100 million) with additional funding of $34.48 (IDA $28.48  million)        Background   When  the  original  project  began,  Nepal  was  facing  serious  political  turmoil  and  uncertainty,  the  result  of  escalating violence associated with longstanding domestic conflict between the government of the time and a  Maoist  insurgency.  Although  the  country  had  seen  improvements  in  the  economy,  its  geographic,  political,  and  social  conditions  brought  considerable  challenges.    The  government  of  Nepal  (GON)  at  that  time  had  begun pushing for more sustained and equitable economic and social development, based on the Tenth Fiveâ€? Year Plan 2002–07.  In 2004, the GoN identified the Poverty Alleviation Fund Project (PAF) as a promising  communityâ€?driven  development  instrument  for  moving  toward  the  government’s  objective  of  social  inclusion, by effectively getting resources to the poorest and most excluded groups in rural Nepal.    The  Project  Development  Objective  was  to  improve  living  conditions,  livelihoods,  and  empowerment  among  the  rural  poor,  with  particular  attention  to  groups  that  have  traditionally  been  excluded  because  of  gender, ethnicity, caste, or location.  Project  Components  were:  (a)  smallâ€?scale  village  and  community  infrastructure;  (b)  income  generating  subprojects; (c) innovation and special programs; (d) capacity building; and (e) administration of PAF II.  The  capacity building component was divided into five subprojects; (1) social mobilization of community groups;  (2) capacity building for local bodies; (3) capacity building for target groups engaged in incomeâ€?generating  activities; (4) support to rural and community finance; and (5) information, monitoring, and evaluation.    Block grant size   With  an  average  of  3,000  people  per  village  development  committee  (VDC),  and  25  VDCs  per  district,  subprojects would target approximately 450,000 individuals or 90,000 households.  The average value of the  subprojects would range from $8,000 to a maximum of $50,000.    Types of Subprojects  At  the  community  level,  PAF  supports  two  types  of  subprojects:  (1)  smallâ€?scale  village  infrastructure  and  services  (microâ€?irrigation,  link  roads,  culverts/bridges,  microâ€?hydro,  water  supply,  sanitation,  schools,  and  health posts); and (2) income generation (agriculture, livestock, cottage industries, trade, and services).           61  INDONESIA: KECAMATAN DEVELOPMENT PROGRAM II   Total Project Period: 1998–2014 12   (Project phase covered by IE: 2003–07)  Total amount for this phase: $421.5 million (IDA $111.3 million)       Background   The  Kecamatan  Development  Project  (KDP)  began  at  a  time  of  significant  upheaval  in  Indonesia’s  history,  when the 1998 economic crisis and political transition left formal government institutions unable to operate.   The political and economic crisis highlighted several key features about development that had been masked  by  the  New  Order  growth  years.  These  included  (a)  despite  high  rates  of  poverty  reduction,  vulnerability  remained high and many poor were not sharing adequately in economic growth; (b) delivery of development  services  to  villages  was  expensive,  of  poor  quality,  and  excessively  centralized;  and  (c)  gaps  between  local  governments and villagers were large.  To help address some of these challenges, KDP facilitated communityâ€? led approaches to service delivery and decisions about the use of local resources.     Project  Development  Objectives  were  to:  (a)  support  participatory  planning  and  development  management in villages; (b) support a broad construction program of social and economic infrastructure in  poor  villages;  and  (c)  strengthen  local  formal  and  informal  institutions  by  making  them  more  inclusive,  accountable, and effective at meeting villagers’ selfâ€?identified development needs.     Project  Components  were:  (a)  Kecamatan  grants  program  ($310.60  million);  (b)  community  capacity  development ($65.5 million); (c) implementation support ($40.4 million); and (c) monitoring and evaluation  studies ($2.9 million).    Block Grant Size  Subdistricts received an annual allocation ranging from $75,000 to $150,000 based on population and on or  off Java.    Types of Subprojects   For  Phase  II,  on  average  more  than  70  percent  of  funded  subprojects  supported  village  infrastructure  and  slightly less than 30 percent went toward revolving funds and social services.                                                                         In 2007, the government of Indonesia scaled up the program and renamed KDP as the National Program for Community  12 Empowerment in Rural Areas (PNPMâ€?Rural).  62  INDONESIA: KECAMATAN DEVELOPMENT PROGRAM BADAN REINTEGRASI ACEH (KDP­BRA)  Total Project Period: 2006–07   (Project phase covered by IE: 2006–07)  Total amount for this phase: $21.7 million       Background   On  August  15,  2005,  the  government  of  Indonesia  (GoI)  and  the  Free  Aceh  Movement  (GAM)  signed  the  Helsinki Memorandum of Understanding (MoU) putting an end to a separatist conflict that lasted nearly 30  years and resulted in almost 30,000 deaths. Civilians frequently suffered the brunt of hostilities.  The tsunami  and the longâ€?standing conflict had devastated the province, its economy, and stateâ€?civic relations.  BRAâ€?KDP  was designed to assist conflict victims across Aceh as part of the reintegration program emerging from the  Helsinki  peace  deal.  Funds  were  channeled  through  the  government’s  Kecamatan  Development  Program  (KDP), which has been operating in Aceh province, and elsewhere in Indonesia since 1998.    The  Project  Development  Objective  was  to  assist  conflictâ€?affected  communities  in  improving  their  living  conditions through the provision of small projects that accord with their needs.      Project Components were:  village block grants, facilitation assistance, and management/M&E costs.    Block Grant Size  All villages within the selected subdistricts received block grants varying from Rp. 60 million (approximately  $6,000) and Rp. 170 million (approximately $ 17,000), with the amount dependent on the intensity of past  conflict in the subdistrict and the population of the village    Types of Subprojects   As  with  KDP,  communities  could  choose  from  an  open  menu  (with  a  negative  list).    Most  beneficiary  communities  preferred  to  use  program  assistance  for  economic  activities  rather  than  common  goods  and  infrastructure. Eightyâ€?nine percent of block grants were spent on livelihood projects, with cattleâ€?breeding and  agriculture inputs the dominant activities. Ten percent of funds were invested in infrastructure.      63  INDONESIA: PNPM – RURAL GENERASI SEHAT DAN CERDAS   Total Project Period: 2007–14 (Project phase covered by IE: 2007–10)  Total amount for this phase: $98 million    Background   Indonesia  has  made  remarkable  strides  in  key  human  development  indicators  over  the  past  few  decades.  Primary  school  enrollment  is  close  to  universal  for  both  boys  and  girls,  and  the  child  mortality  rate  has  declined  rapidly.  Nevertheless,  infant  mortality,  child  malnutrition,  maternal  mortality,  junior  secondary  school enrollment, and educational learning quality have all remained problematic in Indonesia compared to  other  countries  in  the  region.  Furthermore,  achievements  in  these  indicators  reveal  large  geographical  disparities,  with  poorer  outcomes  in  rural  and  remote  provinces  and  districts.  Improving  access  to  basic  quality  health  and  education  services  is  a  key  component  of  an  overall  poverty  reduction  strategy  for  Indonesia.  This pilot project built on the KDP/National Community Empowerment Program CDD platform.    The  Project  Development  Objective  was  to  obtain  12  education  and  health  targets  related  to  improved  school enrollment and attendance and child and maternal health through a community CCT approach.    Project  Components  were:  (a)  performanceâ€?based  community  block  grants  to  support  communities  to  reach 12 education and health indicators; (b) facilitation and training; and (c) implementation support and  technical assistance.    Block Grant Size   Village block grants ranged from an average of $8,500 in 2007 up to an average of $18,200 in 2009.         Types of Subprojects   Block grants were used for education (56 percent) such as school supplies and uniforms, transport, classroom  renovations;  and  health  activities  (44  percent)  such  as  nutritional  feeding  programs,  transport,  medical  equipment, and supplies.                64  INDONESIA: URBAN POVERTY PROJECT ll  Total Project Period: 1999–2011 (Project phase covered by IE: 2002–10)  Total amount for this phase: $126.9 million (IDA $70.5 million, IBRD $29.5 million)       Background   The concept of UPP was developed following the 1997 financial crisis, when the problem of urban poverty in  Indonesia was seen primarily as lack of income and employment.  Midâ€?1998 urban poverty estimates started  at  20  percent.  Out  of  a  91  million  labor  force,  open  urban  unemployment  was  estimated  at  15  million  (16  percent).  There was an urgent need to get financial resources to the poor rapidly.      Project  Development  Objectives  were:  (a)  to  establish  or  support  representative  and  accountable  community  organizations  that  are  able  to  provide  services  to  the  urban  poor  and  increase  the  voice  of  the  poor  in  public  decision  making;  (b)  to  make  local  government  more  responsive  to  the  needs  of  the  poor  through increased cooperation with community organizations; and (c) to improve services for the urban poor  (financial services, social services, and infrastructure).    Project  Components  were:  (a)  community  development  and  local  government  capacity  building  ($17.10  million);  (b)  kelurahan  (urban  village)  grants  ($58.80  million),  which  were  to  provide  villages  with  annual  funds  ranging  from  Rps.  150  million  to  450  million  per  kelurahan,  depending  on  population;  (c)  poverty  alleviation partnership grant ($35.20 million), which allows local governments to access matching grants for  city  level  poverty  reduction  activities  that  are  too  big  to  be  financed  by  the  kelurahan  grants,  and;  (d)  implementation support ($9.70 million).     Block Grant Size  In the kelurahan grants component, each participating kelurahan will have access to a oneâ€?time allocation of  grant funds according to the size of the population, from Rp. 150 million ($15,000) for kelurahans with less  than 3,000 people to Rp 450 million ($45,000) to kelurahans with greater than 10,000 people. To benefit the  poor more, these allocations may be adjusted depending on an estimated number of poor households in the  kelurahan.    The  ceiling  for  any  single  subproject  or  microcredit  per  group  (KSM)  is  Rp.  30  million  ($3,000  equivalent).      Types of Subprojects   In  Phase  I  and  II,  most  communities  had  chosen  a  package  of  services  that  included  small  infrastructure  improvements (roads, drains, water, sanitation, etc.) and revolving funds for providing microâ€?credit services  to community members.  The revolving fund had about 550,000 loans out, of which 60 percent were for small  trade, 15 percent for services, and about 10 percent for manufacturing.      65  PHILIPPINES:  KAPIT­BISIG  LABAN  SA  KAHIRAPAN  –  COMPREHENSIVE  AND  INTEGRATED  DELIVERY  OF SOCIAL SERVICES PROGRAM  Total Project Period: 2002–14  Total amount for the IE phase: $182 million (IBRD $100 million)      Background   Widespread poverty continues to be a challenge in the Philippines.  The 1997 East Asian financial crisis and  the  subsequent  El  Niño  phenomenon  further  exacerbated  the  economic  situation  by  causing  a  reduction  in  the average living standards of Filipinos and an increase in the country’s poverty incidence.  The aggregate  number of households below the poverty line in 2000 reached a high of 5.2 million and absolute poverty was  over 40 percent in 8 of the 16 regions in the country.  KALAHIâ€?CIDSS was set up to fight against poverty by  adopting peopleâ€?centered approaches to problem solving.      The  Project  Development  Objective  was  to  assist  the  borrower  in  strengthening  local  communities'  participation  in  barangay  governance,  and  developing  their  capacity  to  design,  implement,  and  manage  development activities that reduce poverty.    Project  Components  were:  (a)  village  (community)  grants  ($132.20  million)  for  community  development  subprojects,  including  investment  in  economic  and  social  infrastructure,  environmental  conservation  measures,  and  capacity  building;  (b)  implementation  support  ($46.20  million),  which  mobilizes  local  communities  to  participate  and  strengthens  the  capacity  of  local  communities  and  government;  and  (c)  monitoring and evaluation ($2.60 million). (Amounts are for original loan)    Block Grant Size   Under the original loan, block grant sizes ranged from $69,000 (PhP 3.5 million) to a high of $225,000 (PhP  13 million) per year depending on the number of villages in each municipality.  Under the additional financing  arrangement in 2010, the amounts increased to approximately $9,767 multiplied by the number of villages in  each municipality.    Types of Subprojects   Approximately 50 percent of the subprojects were basic social services facilities (e.g., water systems, school  buildings,  health  station  and  day  care  centers),  with  28  percent  for  basic  transport  infrastructure,  such  as  access roads and small bridges.  The rest of the subprojects were community enterprise facilities such as preâ€? post agricultural production facilities (12 percent) and environmental and disaster control infrastructure (10  percent).        66  SENEGAL: PROGRAMME NATIONAL D’INFRASTRUCTURES RURALES (PNIR)  Total Project Period: 2000–12 (Project phase covered by IE: 2000–05)  Total amount for this phase: $42.9 million (IDA $28.5 million)     Background   The 1994 poverty assessment study highlighted the severity of poverty in Senegal, with more than oneâ€? third  of  the  total  population  living  under  the  poverty  line.  Moreover,  poverty  remained  an  essentially  rural  phenomenon,  with  80  percent  of  the  poor  households  living  in  rural  areas.  High  underemployment  was  prevalent  in  rural  areas,  fueling  the  rural  outâ€?migration  of  the  younger  generations.  The  1996  government  reforms, with their primary focus on poverty reduction, strengthened decentralization as a means to promote  socioeconomic development in Senegal and, to that end, transferred a number of sectoral competencies and  resources to local governments.  Under the leadership of the National Council for the Development of Local  Communities,  different  approaches  in  rural  development  were  to  be  unified  to  develop  the  National  Decentralized Rural Development Program.     The  Project  Development  Objectives  were  to:  (a)  improve  decentralized  local  governance  and  local  capacity; (b) establish participatory and decentralized mechanisms for selecting, funding, and implementing  rural community investment programs; (c) strengthen the national institutions supporting decentralization;  and (d) implement basic infrastructure in a selected number of communities.    Project Components were: (1) support for decentralized rural development ($10.7 million) to support and  strengthen  the  institutional,  financial,  and  human  resource  capacity  of  rural  local  governments;  (b)  local  investment fund (LIF) ($13.3 million), which was the mechanism for transferring financial resources to local  governments to fund priority infrastructure spillovers (health and maternity posts, schools, water supply and  sanitation,  culverts,  etc)  identified  in  their  local  development  plans  and  annual  investment  programs.;  (c)  community  roads  program  ($11.6  million);  and  (d)  project  coordination,  monitoring,  and  evaluation  ($6.0  million).     Block Grant Size   It was initially estimated that through the LIF, rural communities would receive on average $70,000 (about  $4 to $5 per inhabitant) annually.  These additional resources were expected to provide funding for two or  three  spillovers,  each  averaging  between  $15,000  and  $20,000.    Beneficiaries  (local/rural  communities)  would be required to contribute at least 20 percent of the total cost in materials, labor, and cash.    Types of Subprojects   There were two types of financing windows in the LIF component: the rural investment fund (RIF) represents  80 percent and would finance basic public infrastructure; and the innovation support fund (ISF) that would  be  an  actionâ€?research  tool  for  a  slightly  larger  menu  of  investment  options,  including  productive  infrastructure.  By the end of the project, 850 microâ€?projects had been implemented, of which 677 were for  the RIF and 183 were under the ISF. This included 298 water supply projects, 141 schools, 140 health centers  and  health  posts,  114  livestock  projects,  60  projects  for  the  promotion  of  women’s  activities,  34  for  agriculture, 21 for youth and sports, 24 for commerce, and 18 miscellaneous projects.  67  SIERRA LEONE: GOBIFO  Total Project Period: 2005–10 (Project phase covered by IE: 2005–09)   Total amount for this phase: $2.5 million financed by Japan Social Development Fund and Italian government.    Background   Though its civil war ended in January 2002, Sierra Leone was still in the process of restoring public services  and rebuilding communities. The conflict destroyed infrastructure, businesses, and local governance capacity  across the country, and by the end of the war, the average citizen survived on the equivalent of 38 cents a day,  and 2 million people, nearly oneâ€? third of the population, had been displaced.  Five years after the end of the  conflict, Sierra Leone was consolidating its return to democratic rule by implementing a decentralization plan  to  empower  locally  elected  district  and  town  councils.    The  GoBifo  project  was  part  of  the  government  of  Sierra  Leone’s  broader  reform  agenda  to  strengthen  participatory  democracy  and  decentralize  public  services.     The  Project  Development  Objectives  were:  (a)  to  strengthen  social  capital  by  enhancing  the  capacity  of  villages  and  local  governments  to  design  and  implement  strategic  development  plans  at  village  and  ward  levels;  and  (b)  to  enhance  accountability  of  decision  makers  through  mechanisms  that  foster  open  and  transparent governance.    Project  Components  were:  (a)  capacity  development  for  village  and  ward  planning;  (b)  capacity  development  to  manage  block  grants  and  communityâ€?driven  initiatives;  and  (c)  capacity  development  for  monitoring and evaluation.    Block Grant Size   About a half of the total project budget was dedicated to villageâ€? and wardâ€?level block grants ($896,000 or 47  percent).  The project disbursed grants worth on average $4,667 to communities with 50 households, or 300  residents (roughly $100 per household, or $16 per capita over three and a half years).      Types of Subprojects   Approximately 300 projects were implemented in the four districts.  Typical projects selected included: (a)  local public goods construction, like schools, road rehabilitation, latrines, community stores, and community  centers;  (b)  agricultural  production  and  livestock  management,  for  example  communal  farms,  grain  drying  floors, and goat herding; (c) skills training and income generating activities, like soap making and carpentry;  and (d) social projects, for example youth football clubs and equipment.     68  ZAMBIA: SOCIAL RECOVERY PROJECT II  Total Project Period: 1991–2005 (Project phase covered by IE: 1995–2000)  Total amount for this phase: $47.7 million (IDA $30 million)    Background   In  1991,  the  country  was  suffering  from  a  stagnant  economy,  poor  health  and  education  indicators,  a  devastating  HIV/AIDS  epidemic,  and  high  poverty  rates.  The  standard  of  living  of  most  Zambians  had  deteriorated sharply since its peak in the early 1970s.  A new government was elected in 1991 and committed  itself to an ambitious economic reform program, which along with macroeconomic reforms also recognized  that it must relieve the social problems that accompanied the economic decline and lessen the impact on the  most  vulnerable  segments  of  the  population.  The  government  of  Zambia  (GoZ)  also  adopted  a  decentralization  policy.      This  institutional  change  would  facilitate  participatory  bottomâ€?up  planning  and  provision of a structure at local levels through which communities could formulate proposals for submission  to different and  appropriate funding sources. In 1991, the Social Recovery Fund was launched supported by  the GoZ and the World Bank.      The  Project  Development  Objective  was  to  assist  the  government’s  poverty  reduction  program  through  financing  of  community  initiatives  and  building  capacity  for  both  quantitative  and  qualitative  poverty  analysis and monitoring.      Project  Components  were:  (a)  community  initiatives  (75  percent  of  total  project  costs),  which  would  finance  spillovers  prepared  and  implemented  by  communities  with  appraisal  and  supervision  from  the  microprojects  unit  (MPU).  A  subâ€?component  would  finance  capacity  building  in  spillovers  implementation  and  decentralized  government  agencies;  (2)  Poverty  Monitoring  (5.7%  of  total  project  cost),  which  would  finance  capacity  building  and  data  collection  through  a  Living  Conditions  Monitoring  Survey  (LCMS)  and  a  systematic participatory monitoring and assessment system; (3) Poverty Analysis (2.3% of total project cost),  which would finance policy relevant research on poverty and social issues in a demandâ€?driven framework.    Block Grant Size  The average cost per microâ€?project was estimated around US$ 75,000.          Types of Subprojects   The project financed small, simple and locally generated community initiatives in education, health, nutrition  and economic infrastructure.   Among the microprojects financed by the project, about 80% were related to  education and about 12% were related to health.    69  BOLIVIA: SOCIAL INVESTMENT FUND II  Total Project Period: 1987–2004 13 (Project phase covered by IE: 1993–98)  Total amount for this phase: $70 million (IDA $40 million)     Background  In  the  light  of  severe  poverty  and  weak  institutional  capacity,  the  government  of  Bolivia  took  emergency  measures to cushion the impact of its stabilization policies in the midâ€?1980s—and  subsequently to address  longer  term  poverty  alleviation—through  a  policy  and  institutional  initiative  such  as  the  Emergency  Social  Fund  (ESF).    ESF  financed  smallâ€?scale  subprojects  in  productive  and  social  infrastructure,  and  of  nonâ€? investment  interventions  in  education,  health,  and  nutrition.    Drawing  on  the  ESF  experiences,  the  Social  Investment  Fund  was  created  to  attract  and  efficiently  channel  external  funding  to  address  country’s  basic  needs in health, water supply, sanitation and education.       The  Project  Development  Objective  was  to  continue  assisting  the  government  to  improve  coverage  and  quality  of  health,  water  supply,  sanitation,  and  education  services  targeting  primarily  the  rural  poor,  including indigenous peoples, disadvantaged women, and children.    Project  Components  were:  (a)  subprojects  ($53.9  million)  in  the  health,  water  supply,  sanitation,  and  education sectors, including subprojects for institutional support to agencies requesting subprojects; (b) an  institutional  development  component,  which  included  administrative  support  to  SIF  (salaries,  training,  and  travel expenses for SIF's staff for $13.3 million); (c) vehicles, office equipment, computers, and miscellaneous  expenses ($1.8 million); and (d) technical assistance, including studies ($1.0 million).    Block grant size     The subproject cost ceiling was $250,000 and the average project cost was estimated at $85,000.     Types of Subprojects  The focus of the subprojects was on the health, water supply, sanitation, and education sectors. Among the  total 2,238 projects, water was 639 (28.6 percent), schools with multiâ€?grade classrooms 563 (25.2 percent),  other  schools  478  (21.4  percent),  health  473  (21.1  percent),  and  sanitation  85  (3.8  percent).    Education  subprojects involved either repairing existing schools or constructing new schools, usually accompanied by  the provision of new desks, blackboards, and playgrounds.  Health subprojects consisted of repairing existing  health centers and constructing new facilities.  SIF water supply investments consisted of providing financing  for smallâ€?scale potable water systems.                                                                       13 According to ICR, the changes made to the system are being supported by IDA and other donors through the “Decentralization  Programmatic Structural Adjustment Credit,â€? which continued until 2004.    70  HONDURAS: SOCIAL INVESTMENT FUND III  Total Project Period: 1990–201214 (Project phase covered by IE: 1995–99)  Total amount for this phase: $117.9 million (IDA $27.4 million)    Background   Until  1990,  Honduras’s  development  strategy  was  based  on  import  substitution,  protected  by  high  tariff  barriers  and extensive  government  intervention,  which  resulted  in a  large  and  inefficient  public  sector  and  serious  structural  imbalances  in  the  economy.  In  1990,  the  government  implemented  a  stabilization  and  adjustment  program,  complemented  by  a  series  of  sectoral  reform  programs.  In  1993,  per  capita  GNP  remained low at an estimated $580. Oneâ€?third of the population of 5.5 million lived in extreme poverty, with  80 percent of the poor living in rural areas.  The Honduran Social Investment Fund (FHIS) was established to  respond to the potentially serious impact of the adjustment program on the country’s poor.    It was hoped  that  FHIS  would  generate  emergency  employment,  and  would  protect  and  improve  the  standard  of  living  among the poor.      The  Project  Development  Objectives  were:  (a)  to  continue  to  assist  the  government  in  strengthening  its  poverty  alleviation  efforts  and  in  maintaining  social  cohesion,  while  the  government  attempted  to  regain  fiscal  balance  and  the  line  ministries  strengthened  their  institutional  capacity;  and  (b)  to  support  the  government's  decentralization  strategy,  development  of  the  local  contracting  industry,  the  sustainability  of  subprojects, and the targeting of scarce resources to the poorer communities.     Project Components were: (a) investment in smallâ€?scale social and economic infrastructure subprojects; (b)  a  priority  “basic  needsâ€?  program  to  support  programs  for  vulnerable  and  disadvantaged  groups;  (c)  a  comprehensive program of technical assistance to help FHIS overcome some institutional weaknesses; (d) a  social data mapping system and strengthening of the design and implementation of the national household  survey as the major tools to target the poor and to monitor progress of social development programs; and (e)  a small program of technical assistance to municipalities in assessing environmental issues and in preparing  programs for corrective action.    Block Grant Size   Over  the  period  1995–98,  FHIS  invested  $96.3  million  in  4,120  smallâ€?scale  infrastructure  projects  in  education,  health,  water  and  sanitation,  and  municipal  improvement.  The  per  project  average  cost  was  approximately $23,374.     Types of Subprojects   As  indicated  above,  the  FHIS  invested  in  smallâ€?scale  infrastructure,  furniture,  and  equipment;  training  for  education; the health and water and sanitation sector; orphanages, child care, and elderly care centers; and  environmental improvement projects.                                                                        The Honduras Social Investment Fund (FHIS) began implementing the Barrio Ciudad Project from 2006–12, focusing on urban  poor  14 areas and municipalities. The project is also supported by the WB.  71  NICARAGUA: SOCIAL INVESTMENT FUND PROJECT II                                                                                      Total Project Period: 1992–200615 (Project phase covered by IE: 1996–98)  Total amount for this phase: $172 million (IDA $26.5 million)    Background  The  government  of  Nicaragua  established  FISE  in  1990  as  a  temporary  social  safety  net  program  to  help  protect  and  improve  the  standard  of  living  of  marginal  groups  during  the  period  of  economic  adjustment,  through the provision of employment and social and economic infrastructure and services.  It was estimated  then  that  oneâ€?half  of  the  country's  population  fell  below  the  poverty  line,  and  oneâ€?fifth  below  the  extreme  poverty  line.    The  government  continued  to  rely  on  FISE  for  the  delivery  of  education,  health,  and  water  supply and sanitation infrastructure to the poorest regions of the country, in line with its social sector policy,  which aimed to expand access to social services, improve efficiency and accountability in the delivery of social  services, and focus on the most vulnerable groups suffering from malnutrition and poverty.     The  Project  Development  Objectives  were  to  help  the  government  sustain  its  poverty  alleviation  efforts  and maintain social  cohesion  during the economic adjustment  period.    Additionally,  the  project  would  help  support  the  government's  decentralization  strategy,  strengthen  the  local  contracting  industry,  broaden  the  menu of projects financed by FISE, support the targeting of resources to more depressed areas, and help meet  the  special  needs  of  the  indigenous  population.  The  IDA  credit  was  also  intended  as  a  channel  to  mobilize  further international support.    Project Components were: (a) investments subprojects ($100 million, of which $10.4 million would be for  FISE's  operating  costs);  (b)  institutional  strengthening  ($1.0  million);  (c)  a  living  standards  measurement  survey ($0.5 million); and (d) a maintenance fund ($0.9 million).    Block grant size and frequency   Subproject  disbursements  were  $85.7  million.    Between  1991  and  1998,  the  total  amount  spent  for  subprojects  was  $174.96  million  for  5,781  contracted  subprojects.  The  average  cost  of  subprojects  varied  depending  on  sectors:  social  infrastructure  ($49,269),  economic  infrastructure  ($93,195),  environment  ($88,721), and social assistance ($8,479).      Types of Subprojects   From  January  1995  through  June  1998,  FISE's  actual  investments  based  on  contracted  subprojects  were:  social infrastructure (66.9 percent), primarily consisting of education (73 percent) and health (13 percent);  plus economic infrastructure (17.4 percent), social services (14.8 percent), and environment (0.9 percent).                                                                        15 FISE III has been followed up by the IDAâ€?financed Poverty Reduction and Local Development Project (2002–06).  72  PERU: SOCIAL FUND (FONCODES)    Total Project Period: 1993–2000 (Project phase covered by IE: 1993–97)  Total amount for this phase: $430 million (IDA $150 million)     Background   A comparison of social indicators among countries in the region showed that Peru ranks near the lowest in  most povertyâ€?related areas such as health, and falls below countries with similar per capita GDPs.  In 1991,  Peru had a GNP per capita of $1,070 equivalent, and an infant mortality of 53 per 1,000 births, higher than  other middleâ€?income countries in Latin America.  The government of Peru (GoP) created FONCODES in 1991  to mitigate the effects of the previous economic crisis and the social cost of economic adjustment through the  financing  of  smallâ€?scale,  shortâ€?term  investment  projects  of  immediate  impact  on  the  poor  and  most  vulnerable.      The Project Development Objectives were to assist the borrower in (a) sustaining its poverty alleviation  efforts;  (b)  mitigating  the  social  costs  of  its  macroeconomic  adjustment;  and  (c)  strengthening  the  institutional  capabilities  of  FONCODES.    In  conjunction  with  these  objectives,  the  project  sought  to  (i)  generate employment and improve access to social services; (ii) promote strong and selfâ€?reliant community  organizations;  (iii)  complement  other  nationwide  projects  in  education  and  health;  and  (iv)  maximize  support from other potential donors.     Project  Components  were:  (a)  communityâ€?based  subprojects  (representing  95.5  percent  of  project  costs),  which supported the financing and supervision of a range of communityâ€?based, laborâ€?intensive subprojects,  with  a  strong  focus on  education,  health,  sanitation,  and  nutrition  along  with  other  activities  to  assist  poor  rural and periâ€?urban populations; and (b) institutional strengthening to improve the institutional capabilities  of  FONCODES  (4.5  percent  of  project  costs).  This  was  to  support  the  rehabilitation  and  refurbishing  of  FONCODES’s facilities, training of staff in project management, organizational analysis for addressing poverty  issues,  and  technical  assistance  related  to  the  implementation  and  supervision  of  the  project,  including  exâ€? post  evaluations  of  completed  subprojects  and  support  for  the  preparation  and  implementation  of  subprojects.      Block Grant Size   Over the course of the project, a total of 12,630 subprojects for $520 million were approved.  The ceiling for  direct contracting with communities of $60,000 was raised midway through the project.  The average cost of  subprojects financed by the Bank loan was about $27,000.      Types of Subprojects  About  75  percent  of  projects  financed  from  1991  to  1995  were  social  infrastructure,  the  vast  majority  of  which  were  primary  schools,  water,  and  sanitation;  7  percent  were  social  assistance;  15  percent  were  economic  infrastructure,  including  small  rural  roads,  bridges,  electricity,  and  irrigation  systems;  and  3  percent  were  productive  projects.  The  loan  financed  the  construction  or  rehabilitation  of  more  than  2,000  primary schools, 1,300 water and sewerage systems, 340 rural roads, 300 irrigation systems, and a variety of  other efforts.   73  ARMENIA: SOCIAL INVESTMENT FUND    Total Project Period: 1996–2012 (Project phase covered by IE: 1996–2000)  Total amount for this phase: $20 million (IDA $12 million)    Background  Facing  immense  development  challenges  stemming  from  the  breakup  of  the  Soviet  system,  the  transition  from a centrally planned to a market economy, a devastating earthquake in 1988, and war and civil disorder  in  bordering  states,  the  government  of  Armenia  (GoA)  aimed  at  achieving  macroeconomic  stability,  promoting  sustainable  growth,  creating  a  strong  private  sector  within  the  framework  of  a  marketâ€?oriented  economy, improving efficiency of the public sector, and protecting vulnerable groups among the population.   In  response  to  these  challenges,  GoA  created  the  Armenian  Social  Investment  Fund  as  an  autonomous  and  demandâ€?driven  agency  to  finance  smallâ€?scale  projects  for  the  rehabilitation  of  basic  social  and  economic  infrastructure.     The Project Development Objective was to support lower income groups among the Armenian population  through  improvement  of  basic  social  services  and  creation  of  employment  opportunities.    The  primary  objectives of the project were  to (a) rehabilitate basic smallâ€?scale infrastructure that can result in immediate  improvements  in  the  living  conditions  of  the  poorest  among  the  population;  (b)  strengthen  the  capacity  of  private smallâ€?scale contractors and other microâ€?businesses so that they can benefit directly or indirectly from  contracts financed by the ASIF; (c) generate employment through financing of laborâ€?intensive public works;  and (d) build greater capacity for policy makers to monitor and analyze trends in the level and structure of  poverty in Armenia.    Project Components were: (a) financing of ASIF microâ€?projects to support the rehabilitation of smallâ€?scale  infrastructure,  including  rehabilitation  of  local  clinics,  primary  schools,  sanitation  work  and  water  supply,  energy  saving  repairs,  repair  of  feeder  roads,  as  well  as  rehabilitation  of  degraded  village  and  town  environments through cleanup and tree planting;  (b) institutional support for the ASIF; (c) capacity building  of  small  contractors,  implementing  agencies,  and  communities;  and  (d)  support  for  monitoring  living  conditions.     Block Grant Size   Since the project began in January 1996, 259 microâ€?projects were completed (as of 2001).  The approximate  average cost of each microâ€?project was $50,000.  However, it was roughly $30,000 during the initial project  years, while during the last years it increased to $75,000.      Types of Subprojects   Of the total 259 microâ€?projects completed under ASIF, 35 percent were smallâ€?scale school rehabilitations, 32  percent  were  potable  water  projects,  11  percent  were  minor  irrigation  works,  and  5  percent  were  health  facilities.  The  remaining  17  percent  of  projects  included  community  centers,  pension  homes,  sewage  and  waste, roads, landscaping, and other.  74  BIBLIOGRAPHY     Alatas,  Vivi.  2005.  “An  Analysis  of  the  Kecamatan  Development  Programâ€?.  Mimeograph.  Jakarta:  The  World  Bank.    Asianâ€?Pacific  Policy  Center.  2007.  Tracking  Progress  Towards  Community  Empowerment  and  Welfare.   Midterm  Survey  for  the  Impact  Evaluation  of  the  Kalahi­CIDSS.  Final  Report.  Manila:  Asianâ€?Pacific  Policy Center.    _______. 2011. Final Survey for the Kalahi­CIDSS Impact Evaluation. Manila: Asiaâ€?Pacific Policy Center.    Araujo,  Caridad  M.,  H.G.  Ferreira  Francisco,  Peter  Lanjouw,  and  Berk  Özler.  2007.  “Local  Inequality  and  Project  Choice:  Theory  and  Evidence  from  Ecuador.â€?  Journal  of  Public  Economics  92  (2008):1022– 1046.    Arcand, Jeanâ€?Louis, Adama Bah, and Julien Labonne. 2011. “Conflict, Ideology and Foreign Aid.â€? Geneva: The  Graduate Institute; CERDI Universite d'Auvergne, and Oxford University.    Arcand,  Jeanâ€?Louis,  and  Leandre  Bassole.  2007.    “Does  Community  Driven  Development  Work?    Evidence  from Senegalâ€?. Draft. CERDIâ€?CNRS, Université d’Auvergne, and EUDN.   Baird,  Sarah,  Craig  McIntosch,  and  Berk  Özler.  2009.  The  Squeaky  Wheels  Get  the  Grease:    Applications  and  Targeting in Tanzania’s Social Action Fund. Washington, DC: The World Bank.    Barrera-Osorio, F., Tazeen Fasih, and Harry A. Patrinos. 2009. Decentralized  Decision­Making  in  Schools.  The  Theory and Evidence on School­Based Management. Washington, DC: The World Bank.   Barron,  Patrick.  2011.  “Communityâ€?Driven  Development  in  Postâ€?Conflict  and  Conflictâ€?Affected  Areas.  Experience  from  East  Asia.â€?  Background  Paper  for  2011  World  Development  Report.  Discussion  Papers. Sustainable Development Department, East Asia and Pacific, World Bank.    Barron, Patrick, Rachael Diprose, and Michael Woolcock. 2006. “Local Conflict and Community Development  in  Indonesia:  Assessing  the  Impact  of  the  Kecamatan  Development  Program.â€?  Indonesian  Social  Development Paper No.10. Washington, DC: World Bank.    Barron,  Patrick,  Macartan  Humphreys,  Laura  Paler,  and  Jeremy  Weinstein.  2009.  “Communityâ€?Based  Reintegration  in  Aceh:  Assessing  the  Impacts  of  BRAâ€?KDP.â€?  Indonesian  Social  Development  Paper  No.12.Washington, DC:  World Bank.    Beath, Andrew, Fotini Christia, Reuben Enikolopov, and Shahim Ahmad Kabuli. 2010.   “Randomized Impact  Evaluation  of  Phase  II  of  Afghanistan’s  National  Solidarity  Programme  (NSP).  Estimates  of  Interim  Program Impact from First Followâ€?up Survey. NSP, Afghanistan.    75  Binswanger,  Hans,  Fátima    Amazonas,  Túlio  Barbosa,  Alberto  Costa,  Naércio  Menezes,    Elaine  Pazello,  and  Claudia  Romano.    2009.    Rural  Poverty  Reduction  in  Northeast  Brazil.    An  Evaluation  of  Community  Driven Development. Washington, DC: World Bank.    Bruns,  Barbara,  Deon  Filmer,  and  Harry  A.  Patrinos.  2011.  Making  Schools  Work:  New  Evidence  on  Accountability Reforms. Washington, DC: World Bank.    Casey,  Katherine,  Rachel  Glennerster,  and  Edward  Miguel.    2011a.  “The  GoBifo  Project  Evaluation  Report:  Assessing  the  Impacts  of  Community  Driven  Development  in  Sierra  Leone.â€?  Final  Report.  Brown  University, Abdul Latif Jameel Poverty Action Lab, and University of California, Berkeley.    _____.  2011b.  “Reshaping  Institutions:  Evidence  on  Aid  Impacts  Using  a  Preâ€?Analysis  Plan.â€?  Draft.  Brown  University,  Abdul Latif Jameel Poverty Action Lab, and University of California, Berkeley.    Chase, Robert S. 2002. “Impact Evaluation of Social Funds. Supporting Communities in Transition: The Impact  of the Armenian Social Investment Fund.â€? World Bank Economic Review 16 (2): 219–240.    Chase,  Robert  S.,  and  Lynne  Sherburneâ€?Benz.  2001.    Household  Effects  of  African  Community  Initiatives:  Evaluating the Impact of the Zambia Social Fund. Washington, DC: World Bank.    Chase, Robert S., and Camilla R. Holmemo.  2005. CDD and Social Capital Impact: Designing a Baseline Survey  in the Philippines. Social Development Department. Washington, DC:  World Bank.    Chase,  Robert  S.,  Rikke  Nording  Christensen,    and  Maniemai  Thongyou.  2006.  Picking  Winners  or  Making  Them? Evaluating the Social Capital Impact of CDD in Thailand. Washington, DC: World Bank.    Coady,  David,  Xinyi  Dai,  and  Limin  Wang.  2001.  “Community  Programs  and  Women’s  Participation:  The  Chinese Experience.â€? Policy Research Working Paper No.2622. East Asia and Pacific Region, Poverty  Reduction and Economic Management Sector Unit. Washington, DC: World Bank.    Coirolo, Luis, and Jill Lammert. 2009.  Rural Poverty Reduction in Northeast Brazil.  Achieving Results through  Community Driven Development. Vol. 1. Washington, DC: World Bank.    Crost,  Benjamin,  and  Patrick  B.  Johnston.  2010.  Aid  Under  Fire:  Development  Projects  and  Civil  Conflict.  Cambridge, MA: Havard Kennedy School, Belfer Center for Science and International Affairs.  P De  Castillo,  Cecilia  Perez.  1998.  “Economic  Analysis  of  Social  Investment  Fund  Projects.â€?  Washington,  DC:  World Bank.    De Janvry, Alain,  Hideyuki Nakagawa, and Elisabeth Sadoulet. 2009. “Proâ€?poor targeting   and electoral rewards in decentralizing to communities the provision of local public   goods in rural Zambiaâ€?. Berkeley: University of California at Berkeley, 29.     Deininger, Klaus, and Yanyan Liu. 2009. “Longerâ€?Term Economic Impacts of Selfâ€?Help Groups  in  India.â€?  Policy  Research  Working  Paper  No.4886.  Development  Research  Group,  Sustainable    Rural  and  Urban Development Team. Washington, DC: World Bank.  76   De Silva, Samantha, and Juneâ€?Wei Sum. 2008. “Social Funds as an Instrument of Social Protection: An Analysis  of  Lending  Trends.  FY2000–2007.â€?  SP  Discussion  Paper  No.0809.  Social  Protection  &  Labor.  Washington, DC: World Bank.    ESA Consultores International. 2004. “St Lucia Poverty Reduction Fund: 2003 Impact   Evaluation Survey Final Report.â€?     _____. 2005. “Evaluación Exâ€?post Del Fondo Hondureño de Inversión Social (FHIS 3).â€? Elaborado a solicitud del  FHIS. Informe Borrador. Tegucigalpa.    Galab,  S.,  and  P.  Prudhvikar  Reddy.  2010.  Impact  Assessment  of  Andhra  Pradesh  Rural  Poverty  Reduction  Project. Centre for Economic and Social Studies, Andhra Pradesh.    Grootaert,  Christiaan,  and  Thierry  van  Bastelaer.  2002.  Understanding  and  Measuring  Social  Capital:  A  Multidisciplinary Tool for Practitioners. Washington, DC: World Bank.    Gugerty, Mary Kay, and Michael Kremer. 2006. Outside Funding and the Dynamics of  Participation  in  Community  Associations.  Seattle,    WA,  and  Cambridge,  MA:  University  of    Washington  and  Harvard University.    Instituto  Apoyo.  2000.  “Sexta  Evaluación  ExPost  del  Foncodes.  Evaluación  de  Impacto  y  Sostenibilidadâ€?.  Informe Final.    Jimat Consult P/L (Zimbabwe), in association with ITAD Limited (UK) and O&M Associates   (Malawi). 2008. “MASAF 3 APL 1 Impact Evaluation Final Report.â€?    Kishindo, Paul.  2000. “Community Project Funding in Malawi under the Malawi Social Action Fund (MSAF)  Demandâ€?Driven Approach:  Potential for Perpetuating Imbalances in Development.â€? Journal of Social  Development in Africa 15(1): 5–14.    Koolwal, Gayatri B. 2005. Is Group Heterogeneity Better? Evidence from Women’s Community   Network Participation in India. Washington, DC: World Bank.    Labonne, Julien.  2011. The Kalahi­CIDSS Impact Evaluation: A Synthesis Report. Discussion Paper. Sustainable  Development, East Asia and Pacific Region.  Washington DC: World Bank.    Labonne,  Julien  and  Robert  S.  Chase.  2008.  Do  Community­Driven  Development  Projects  Enhance  Social  Capital?  Evidence  from  the  Philippines.  Sustainable  Development  Network,  Social  Development  Department. Washington, DC: World Bank.    Labonne, Julien and Robert S. Chase.  2009.  “Who is at the Wheel When Communities Drive     Development?  Evidence from the Philippines.â€?  World Development  37, 219â€?231.    Lokshin, Michael and Ruslan Yemtsov. 2004.  “Evaluating the Impact of Infrastructure Rehabilitation Projects  on Household Welfare in rural Georgia.â€? Washington, DC: World Bank.    77  Mansuri,  Ghazala,  and  Vijayendra  Rao.  2004.  “Communityâ€?Based  and  –  Driven  Development:  A  Critical  Reviewâ€? World Bank Researcher Observer, 19, 1â€?39.    ______.  2011.  “Localizing  Development:  Does  Participation  Work?â€?  World  Bank  Policy  Research  Report.  (forthcoming). Washington, DC: World Bank.    Morel,  Adrian,  Makiko  Watanabe,  and  Robert  Wrobel.  2009.  Delivering  Assistance  to  Conflict­Affected  Communities: The BRA­KDP Program in Aceh. Jakarta: World Bank.    Newman,  John,  Menno  Pradhan,  Laura  B.  Rawlings,  Geert  Ridder,  Ramiro  Coa,  and  Jose  Luis  Evia.  2000.   “Impact  Evaluation  of  Social  Funds.  An  Impact  Evaluation  of  Education,  Health,  and  Water  Supply  Investments by the Bolivian Social Investment Fund.â€? The World Bank Economic Review 16 (2): 241– 274.    Olken, Benjamin A., Junko Onishi, and Susan Wong. 2010. Indonesia’s PNPM Generasi Program: Interim Impact  Evaluation Report. Jakarta: World Bank.    _____. 2011a. Indonesia’s PNPM Generasi Program: Final Impact Evaluation Report. Jakarta: World Bank.    _____. 2011b.  “Should Aid Reward Performance?  Evidence from a Field Experiment on Health and Education  in Indonesiaâ€?. NBER Working Paper 17892.      Parajuli, Dilip, Gayatri Acharya, Nazmul Chaudhury, and Thapa Bishnu Bahadur. 2012.  “Impact of Social Fund  on  the  Welfare  of  Rural  Households:  Evidence  from  Nepal  Poverty  Action  Fund.â€?  Washington,  DC:  World Bank.    Paxson,  Christina,  and  Norbert  R.  Schady.  2002.  “Impact  Evaluation  of  Social  Funds.  The  Allocation  and  Impact of Social Funds; Spending on School Infrastructure in Peru.â€? World Bank Economic Review 16  (2): 297–319.    Pradhan,  Menno,  Vijayendra  Rao  and  Christina  Rosemberg.  2010.  “The  Impact  of  the  Community  Level  Activities of the Second Urban Poverty Project.â€? Draft.    Rao,  Vijayendra,  and  Ana  Maria  Ibañez.    2003.    “The  Social  Impact  of  Social  Funds  in  Jamaica.    A  Mixedâ€? Methods  Analysis  of  Participation,  Targeting,  and  Collective  Action  in  Communityâ€?Driven  Development.â€?  Policy  Research  Working  Paper  No.2970.  Development  Research  Group,  Poverty  Team. Washington, DC: World Bank.    Rawlings,  Laura,  Lynne  Sherburneâ€?Benz,  and  Julie  van  Domelen.  2003.  Evaluating  Social  Funds:  A  Cross­ Country Analysis of Community Investments. Washington, DC: World Bank    Schady, Norbert R. 2000. “The Political Economy of Expenditures by the Peruvian Social Fund (FONCODES),  1991–95.â€? The American Political Science Review 94(2): 289–304.     Serranoâ€?Berthet,  Rodrigo.  2007.  “From  Social  Funds  to  Local  Governance  and  Social  Inclusion  Programs.  A  Prospective Review from the ECA Region.â€?  Volume I: Main Report. Washington, DC: World Bank.  78  _____.  2007.  “From  Social  Funds  to  Local  Governance  and  Social  Inclusion  Programs.    A  Prospective  Review  from the ECA Region.â€? Volume II: Technical Annexes. Washington, DC: World Bank.     _____. 2008.  “Social and Local Development funds in the Africa Region.  Evolution and Options.â€? Prepared for  HD Learning Week, LCSSO. Washington, DC: World Bank.    Singh,  Surjit,  and  Radheyshyam  Sharma.    2004.    Impact  Assessment  of  the  District  Poverty  Initiative  Project  Rajasthan. Jaipur: Institute of Development Studies.    Van Domelen, Julie. 2003. Social Capital in the Operations and Impacts of Social Investment Funds. Washington,  DC: World Bank.    Voss,  John.  2008.  Impact  Evaluation  of  the  Second  Phase  of  the  Kecamatan  Development  Program.  Jakarta:  World Bank.    Wassenich,  Paul,  and  Katherine  Whiteside.  2003.  “CDD  Impact  Assessments  Study:  Optimizing  Evaluation  Design under Constraints.â€?  Washington, DC: World Bank.    Wong, Susan. 2003. Indonesia Kecamatan Development Program: Building a Monitoring and Evaluation System  for a Large­Scale Community­Driven Development Program. Discussion Paper. Environment and Social  Development Unit, East Asia and Pacific Region. Washington, DC: World Bank.    World Bank.1992. “Nicaragua.â€? Social Investment Fund Project. Staff Appraisal Report. Report No.11108â€?NI.  Country  Department  II,  Human  Resources  Operations  Division,  Latin  America  and  the  Caribbean  Regional Office.     _____.  1993a.  “Bolivia.â€?  Second  Social  Investment Fund  Project.  Staff  Appraisal  Report.  Report  No.11927â€?BO.  Human Resources Division, Country Department III, Latin America and the Caribbean Region.    _____.  1993b.  “Peru.â€?  Social  Development  and  Compensation  Fund  (FONCODES)  Project.  Staff  Appraisal  Report.  Report  No.12245â€?PE.  Latin  America  and  the  Caribbean  Regional  Office,  Human  Resources  Operations Division, Country Department I.    _____.1995a.  “Armenia.â€?  Social  Investment  Fund  Project.  Staff  Appraisal  Report.  Report  No.14655â€?AM.  Municipal & Social Services Division, Country Department IV, Europe and Central Asia Region.    _____.1995b.  “Honduras.â€?  Third  Social  Investment  Fund  Project  (FHIS  Second  Stage).  Staff  Appraisal  Report.  Report No.14181â€?HO. Country Department II, Human Resources Operations Division, Latin America  and the Caribbean Regional Office.    _____.1995c.  “Zambia.â€?  Second  Social  Recovery  Project.  Staff  Appraisal  Report.  Report  No.14329â€?ZA.  Human  Resources Operations Division, Southern Africa Department, Africa Regional Office.    _____.1997.  “Georgia.â€?  Social  Investment  Fund  Project.  PAD.  Report  No.17144â€?GE.  Infrastructure  Sector,  Europe and Central Asia Regional Office.    79  _____.1998a.  “Nicaragua.â€?  Second  Investment  Fund  Project.  ICR.  Report  No.18657.Human  and  Social  Development  Department  (LCHSD),  Central  Amercia  Country  Management  Unit,  Latin  America  and  the Caribbean Regional Office.    _____.1998b. “Peru.â€? Social Development and Compensation Fund Project (FONCODES). ICR. Report No.18016.  Sector Management Unit, Human Development, Latin America and the Caribbean Regional Office.    _____.1999. “Senegal.â€? National Rural Infrastructure Project In Support of the First Phase of the National Rural  Infrastructure  Program.  PAD.  Report  No.19963â€?SE.  AFTT2/AFTR3,  Country  Department  14,  Africa  Region.    _____.2000a.  “Honduras.â€?  Third  Social  Investment  Fund  Project  (FHIS  Second  Stage).  ICR.  Report  No.19589.  Human  and  Social  Development  Department  (LCHSD),  Central  America  Country  Management  Unit,  Latin America and the Caribbean Regional Office.    _____.2000b.  “India.â€?  Madhya  Pradesh  District  Poverty  Initiatives  Project.  PAD.  Report  No.21020â€?IN.  SASRD,  South Asia Regional Office.    _____.2000c.  “India.â€?  Andhra  Pradesh  District  Poverty  Initiative  Project.  PAD.  Report  No.20089â€?IN.  SASRD,  South Asia Regional Office.    _____.2000d. “Nicaragua.â€? Exâ€?Post Impact Evaluation of the Emergency Social Investment Fund (FISE). Report  No.20400â€?NI.  Human  Development  Sector  Management  Unit,  Latin  America  and  the  Caribbean  Region.    _____.2001a. “Armenia.â€? Social Investment Fund. ICR. Report No.22077.    _____.2001b.  “Bolivia.â€?  Soc  Invest  Fund  II.  ICR.  Report  No.22450.  Country  Management  Unit  for  Bolivia,  Ecuador and Peru. Human Development Sector Management Unit, Latin America and the Caribbean  Region.    _____.2001c. “Indonesia.â€? Second Kecamatan Development Project. PAD. Report No. 22279â€?IND. Environment  and Social Development Unit, East Asia and Pacific Region.    _____.2001d.  “Zambia.â€?  Second  Social  Recovery  Project.  ICR.  Report  No.22047â€?ZA.  Human  Development  1,  Country Department 3, Africa Region.    _____.2002.  “Philippines.â€?  Kapitbisig  Laban  Sa  Kahirapanâ€?Comprehensive  and  Integrated  Delivery  of  Social  Services  (Kalahiâ€?CIDSS)  Report.  Report  No.24642â€?PH.  Environment  and  Social  Development  Unit,  East Asia and Pacific Region.    _____.2006. “Senegal.â€? National Rural Infrastructure Project. ICR. Report No.36541.    _____  2007a.    Enabling  East  Asian  Communities  to  Drive  Local  Development:  East  Asian  CDD  Flagship  Report.   Washington, DC: World Bank.    80  _____.2007b. “Nepal.â€? Poverty Alleviation Fund Project II. PAD. Report No.41317â€?NP. Sustainable Development  Department, Agriculture and Rural Development Unit, South Asia.    _____.2007c.  “India.â€?  Andhra  Pradesh  District  Poverty  Initiatives  Project.  ICR.  Report  No.  ICR000000098.  Sustainable Development Sector Unit, India Country Management Unit, South Asia Region.    _____.2007d. From Social Funds to Local Governance and Social Inclusion Programs. A Prospective Review from  the  ECA  Region.  Volume  I:  Main  Report.  Report  No.39953â€?ECA.  Human  Development  Sector  Unit,  Europe and Central Asia Region.    _____.2007e. From Social Funds to Local Governance and Social Inclusion Programs. A Prospective Review from  the  ECA  Region.  Volume  II:  Technical  Annexes.  Report  No.39953â€?ECA.  Human  Development  Sector  Unit, Europe and Central Asia Region.    _____.2008a.  “India.â€?  Madhya  Pradesh  District  Poverty  Initiatives  Project.  ICR.  Report  No.ICR00001029.  Sustainable  Development  Sector  Unit,  Agriculture  and  Rural  Development  Unit,  India  Country  Management Unit, South Asia.    _____.2008b.  “Indonesia.â€?  Second  Kecamatan  Development  Project.  ICR.  Report  No.ICR0000767.  Sustainable  Development Sector Unit, Indonesia Country Management Unit, East Asia and Pacific Region.    _____.2008c.  “Indonesia.â€?  National  Program  for  Community  Empowerment  in  Rural  Areas.  PAD.  Report  No.42409â€?ID. Sustainable Development Department, East Asia and Pacific Region.    _____.2008d.  “Georgia.â€?  Social  Investment  Fund  and  Supplemental  Emergency  Earthquake  Rehabilitation  Project. ICR. Report No. ICR0000703. Human Development Sector Unit, South Caucasus Country Unit,  Europe and Central Asia Region.    _____.2008e.  Social  and  Local  Development  Funds  in  the  Africa  Region.  Evolution  and  Options.  Human  Development  Sector  Unit,  Africa  Region,  Social  Protection  Department,  Social  Development  Department. (Draft).    _____.2008f.  Indonesia.  Conditional  Cash  Transfers  in  Indonesia:  Baseline  Survey  Report  Program  Keluarga  Harapan and PNPM­Generasi.  Jarkata: World Bank.    _____.2008g.  Social  Development  in  Operations  –  Social  Development  in  Action  across  the  Bank.  Social  Development Department Publication. (Draft)    _____.2009a. “Nepal.â€? Poverty Alleviation Fund Project. Implementation Completion and Results Report (ICR).  Report No.ICR00001197. Sustainable Development Sector Unit, Nepal, South Asia Region.    _____.2009b.  “India.â€?  Second  Madhya  Pradesh  District  Poverty  Initiatives  Project  (MPDPIPâ€?II).  PAD.  Report  No.47448â€?IN.Sustainable  Development  Department,  Agriculture  and  Rural  Development  Unit,  India  Country Management Unit, South Asia Region.    81  _____.2010a.  “Afghanistan.â€?  National  Solidarity  Program  III.  Project  Information  Document  (PID).  Report  No.AB5745. South Asia Region.    _____.2010b.  “Afghanistan.â€?  Third  Emergency  National  Solidarity  Project.  Emergency  Project  Paper.  Report  No.54540â€?AF.  Sustainable  Development  Department,  Agriculture  and  Rural  Development  Sector,  Afghanistan Country Management Unit, South Asia Region.    _____.2010c.  “Indonesia.â€?  Third  National  Program  for  Community  Empowerment  in  Rural  Areas.  Project  Appraisal  Document  (PAD).  Report  No.53046â€?ID.  Indonesia  Sustainable  Development  Department,  East Asia and Pacific Region.    _____.2010d. “Philippines.â€? Kapitbisig Laban Sa Kahirapan â€? Comprehensive and Integrated Delivery of Social  Services Project. Project Paper. Report No.56355â€?PH. Philippines Sustainable Development Unit, East  Asia and Pacific Region.    _____.2010e.  “Sierra  Leone.â€?  Capacity  Development  to  Strengthen  Social  Capital  (GoBifo  Project).  Review  of  Japan  Social  Development  Fund  (JSDF)  Grants  Closed  in  FY08  and  FY09.  Concessional  Finance  and  Global Partnerships (CFP).    _____.2010f. “Sierra Leone.â€? Capacity Development to Strengthen Social Capital (GoBifo Project). GRM. Internal  Bank Report.    _____.  2011a.  Evidence  and  Lessons  Learned  from  Impact  Evaluations  on  Social  Safety  Nets.  Independent  Evaluation Group (IEG).    _____.  2011b.  “Sierra  Leone.â€?  Youth  Participation  and  Empowerment  Project.  Grant  Reporting  Monitoring  (GRM). Internal Bank Report, Social Development Department.    _____.2011c.  “Sierra  Leone.â€?  GoBifoâ€?Strenthening  Social  Capital  through  Capacity  Development.  Event  Brief,  Japan Social Development Fund (JSDF Day).                          82Â